Guo 2015

Design and modelling of optimal
driveline control strategy for an electric

racing car with rear in-line motors
Guo, M.
Submitted version deposited in CURVE March 2016
Original citation:
Guo, M. (2015). Design and modelling of optimal driveline control strategy for an electric
racing car with rear in-line motors. Unpublished PhD Thesis. Coventry: Coventry University
Copyright and Moral Rights are retained by the author. A copy can be downloaded for
personal non-commercial research or study, without prior permission or charge. This item
cannot be reproduced or quoted extensively from without first obtaining permission in
writing from the copyright holder(s). The content must not be changed in any way or sold
commercially in any format or medium without the formal permission of the copyright
holders.
Some materials have been removed from this thesis due to third party copyright. Pages
where material has been removed are clearly marked in the electronic version. The
unabridged version of the thesis can be viewed at the Lanchester Library, Coventry
University.
CURVE is the Institutional Repository for Coventry University

http://curve.coventry.ac.uk/open
DESIGN AND MODELLING OF OPTIMAL DRIVELINE
CONTROL STRATEGY FOR AN ELECTRIC RACING CAR
WITH REAR IN-LINE MOTORS
Meng Guo
A thesis submitted in partial fulfilment of the
requirements of the Coventry University
for the degree of Doctor of Philosophy
August 2015
Abstract
Interest in electric vehicles (EVs) has increased rapidly over recent years from both
industrial and academic viewpoints due to increasing concerns about environmental
pollution and global oil usage. In the automotive sector, huge efforts have been
invested in vehicle technology to improve efficiency and reduce carbon emissions
with, for example, electric vehicles. Nowadays, the safety and handling of electric
vehicles present new tasks for vehicle dynamics engineers due to the changes in
weight distribution and vehicle architecture. This thesis focuses on one design area
of the electric vehicle torque vectoring control with the aim of investigating the
potential benefits of improved vehicle dynamics and handling for EVs.
A full electric racing car kit developed by Westfield Sportcars based on an in-line
motors design has been modelled in ADAMS with typical subsystems, and then
simulated with computer-based kinematic and dynamic analyses. Thus, the
characteristics of the suspensions and the natural frequencies of the sprung and
unsprung masses were found, so that the model was validated for further simulation
and investigation. Different architectures of the EVs, namely the in-line motors and
the in-wheel motors, are compared using objective measurements. The objective
measurements predicted with kinematics, dynamics and handling analyses confirm
that the architecture of the in-line motors provides a superior dynamics performance
for ride and driveability. An Optimal Driveline Control Strategy (ODCS) based on the
concept of individual wheel control is designed and its performance is compared with
the more common driveline used successfully in the past. The research challenge is
to investigate the optimisation of the driving torque outputs to control the vehicle and
provide the desired vehicle dynamics. The simulation results confirm that active yaw
control is indeed achievable.
The original aspects of this work include defining the characteristics and linearity of
the project vehicle using a novel consideration of yaw rate gain; the design and
development the Optimal Driveline Control Strategy (ODCS); the analysis and
modelling the ODCS in the vehicle and the comparison of the results with
conventional drivelines. The work has demonstrated that valuable performance
benefits result from using optimal torque vectoring control for electric vehicle
Contents
Index for Figures ....................................................................................................... IV
Index for tables ........................................................................................................ VIII
Annotations ............................................................................................................... IX
Abbreviations ............................................................................................................. X
Acknowledgement ..................................................................................................... XI
1. Introduction ......................................................................................................... 1
1.1 Pollution and fuel consumption ..................................................................... 1
1.2 Developing electric vehicles .......................................................................... 1
1.3 Aim and objectives ........................................................................................ 3
2 Literature Review ................................................................................................ 6
2.1 Introduction ................................................................................................... 6
2.2 Electric Vehicle.............................................................................................. 6
2.3 Drivetrain Designs ......................................................................................... 7
2.3.1 Conventional mechanical drivetrains ...................................................... 7

2.3.2 Drivetrains for electric vehicle ................................................................. 8
2.4 Vehicle dynamics control ............................................................................ 10
2.4.1 Direct Yaw Control ................................................................................ 11

2.4.2 Torque Vectoring System ..................................................................... 13
2.5 Handling Control for electric vehicles .......................................................... 18
2.5.1 Driver Model ......................................................................................... 22

2.5.2 Importance of control strategy .............................................................. 24
2.6 Concluding remarks .................................................................................... 26
3. Modelling Electric Vehicle in ADAMS ................................................................ 29
3.1 Introduction ................................................................................................. 29
3.2 Project Vehicle ............................................................................................ 29
3.2.1 Driveline................................................................................................ 30
3.2.2 Chassis and Suspension system .......................................................... 32
I
3.3 Modelling project vehicle ............................................................................. 33
3.3.1 Vehicle Body......................................................................................... 34

3.3.2 Suspension System .............................................................................. 36
3.3.3 Steering System ................................................................................... 40
3.3.4 Aerodynamic Effects ............................................................................. 42
3.3.5 Driveline Modelling ............................................................................... 44
3.4 Tyre modelling............................................................................................. 47
3.4.1 Modelling virtual tyre rig ........................................................................ 47

3.4.2 Tyre model for project vehicle ............................................................... 48
3.5 Conclusion .................................................................................................. 51
4 Measurement and analysis of the virtual model ................................................ 53
4.1 Introduction ................................................................................................. 53
4.2 Kinematic analysis ...................................................................................... 53
4.2.1 Quarter vehicle modelling ..................................................................... 54

4.2.2 General approach for kinematic analysis .............................................. 55
4.2.3 Suspension measurements .................................................................. 56
4.3 Dynamic analysis ........................................................................................ 60
4.3.1 Total degrees of freedom...................................................................... 61

4.3.2 Dynamic model ..................................................................................... 62
4.4 Steady-State Handling analysis .................................................................. 68
4.4.1 Driver behaviour modelling a path following controller ..................... 68

4.4.2 Driver behaviour modelling a survey controller ................................ 74
4.4.3 Steering torque input ............................................................................ 75
4.5 Simulation results ........................................................................................ 77
4.5.1 Steady state cornering behaviour ......................................................... 77

4.5.2 Path following behaviour....................................................................... 78
4.5.3 Vehicle linearity .................................................................................... 79
4.6 Conclusion .................................................................................................. 81
5. Comparison of architecture for electric vehicle ................................................. 83
5.1 Introduction ................................................................................................. 83
5.2 Architecture of in-wheel motors vehicle ....................................................... 83
5.3 Modelling and validating of in-wheel motors vehicle ................................... 86
II
5.4 Modelling architecture of in-line motors vehicle........................................... 88
5.5 Simulation Results ...................................................................................... 91
5.3.1 Analysis of ride comfort ........................................................................ 91

5.3.2 Analysis of drivability check ................................................................ 100
5.5 Conclusion ................................................................................................ 107
6 Torque vectoring system with in-line motors ................................................... 109
6.1 Introduction ............................................................................................... 109
6.2 Basic principles ......................................................................................... 110
6.3 Optimal Driveline Control Strategy (ODCS) .............................................. 114
6.3.1 Definitions of Desired Dynamics Behaviour ........................................ 116

6.3.2 Secondary Control .............................................................................. 125
6.3.3 Advanced Torque Vectoring Control (ATVC) ...................................... 129
6.4 Simulation results ...................................................................................... 136
6.4.1 Steady state cornering ........................................................................ 136

6.4.2 Double lane change ............................................................................ 141
6.5 Conclusion ................................................................................................ 147
7. Conclusions and future work ........................................................................... 149
7.1. Summary and conclusions ........................................................................ 149
7.2. Future work ............................................................................................... 153
Bibliography ........................................................................................................... 154
Appendix ................................................................................................................ 161
III
Index for Figures
Figure 2-1 Acceleration cornering performance ...................................................... 13

Figure 2-2 Schematic of the active limited-slip differentialError! Bookmark not
defined.
Figure 2-3 Vectoring torque acting on rear right and left wheels ............................. 15
Figure 2-4 Effect of torque vectoring ........................................................................ 15
Figure 2-5 Left-Right Torque Vectoring Concept ...................................................... 17
Figure 2-6 Schematics of recent torque vectoring systems ...................................... 17
Figure 2-7 Wide lane change at 70km/h (dry asphalt) .............................................. 21
Figure 2-8 Control strategy for brake-based torque vectoring .................................. 24
Figure 2-9 Control structure for torque vectoring .........Error! Bookmark not defined.
Figure 2-10 Schematic diagram of the driving control algorithm .............................. 26
Figure 3-1 Architecture of Westfield i-Racer............................................................. 30
Figure 3-2 The position of the electric motors .......................................................... 31
Figure 3-3 The position of the battery package ........................................................ 31
Figure 3-4 The pictures for the chassis and suspensions ........................................ 32
Figure 3-5 The pictures of the front and rear suspensions ....................................... 32
Figure 3-6 Westfield i-Racer model in ADAMS/View ................................................ 33
Figure 3-7 The method for locating vehicle centre of mass height ........................... 35
Figure 3-8 The actual front and rear suspensions .................................................... 36
Figure 3-9 Front and rear suspension model in ADAMS/View ................................. 37
Figure 3-10 The 2D curve for the nonlinear damper ................................................ 38
Figure 3-11 The parts connected by the correct joints ............................................. 40
Figure 3-12 Modelling steering system in ADAMS ................................................... 41
Figure 3-13 The steering ratio for Westfield i-Racer model ...................................... 42
Figure 3-14 Aerodynamic drag forces in the lateral and longitudinal direction ........ 43
Figure 3-15 The motors position of the project vehicle ............................................. 44
Figure 3-16 Modelling the driveline in ADAMS/View ................................................ 45
Figure 3-17 The tyre test rig model .......................................................................... 48
Figure 3-18 Lateral Force Vs Slip Angle tested on the Virtual Tyre Test Rig ........... 51
Figure 3-19 Self Aligning Moment vs Slip Angle ...................................................... 51
Figure 4-1 Modelling the front of the quarter project vehicle .................................... 54
Figure 4-2 Kinematic simulation for suspension system........................................... 56
IV
Figure 4-3 Half Track Change Vs Bump Movement (Front) ................................... 179
Figure 4-4 Wheel Recession Vs Bump Movement (Front) .................................... 180
Figure 4-5 Steering Axis Inclination Vs Bump Movement (Front) ........................... 180
Figure 4-6 Ground Level Offset Vs Bump Movement (Front) ................................. 180
Figure 4-7 Castor Angle Vs Bump Movement (Front) ............................................ 181
Figure 4-8 Suspension Trail Vs Bump Movement (Front) ...................................... 181
Figure 4-9 Steer Angle Vs Bump Movement (Front) .............................................. 181
Figure 4-10 Half Track Change Vs Bump Movement (Rear) .................................. 182
Figure 4-11 Wheel Recession Vs Bump Movement (Rear) .................................... 182
Figure 4-12 Steer Axis Inclination Vs Bump Movement (Rear) .............................. 183
Figure 4-13 Ground Level Offset Vs Bump Movement (Rear)................................ 183
Figure 4-14 Castor Angle Vs Bump Movement (Rear) ........................................... 183
Figure 4-15 Suspension Trail Vs Bump Movement (Rear) ..................................... 184
Figure 4-16 Camber Angle Vs Bump Movement (Rear) ......................................... 184
Figure 4-17 Steer Angle Vs Bump Movement (Rear) ............................................. 184
Figure 4-18 Two degrees of freedom quarter vehicle model .................................... 60
Figure 4-19 The dynamic model for modal solution ................................................. 63
Figure 4-20 Animations of the modes at primary ride behaviour .............................. 66
Figure 4-21 Animations of the modes at wheel hop behaviour................................. 67
Figure 4-22 Explanation of Ground Plane Velocity ................................................... 70
Figure 4-23 Explanation of Demanded Yaw Rate .................................................... 71
Figure 4-24 Path curvature for steady state cornering ............................................. 72
Figure 4-25 Steering torque acting on steering wheel .............................................. 76
Figure 4-26 Steering input for the path following model.......................................... 76
Figure 4-27 The explanation of Under-steer or Over-steer....................................... 77
Figure 4-28 Determination of the project model ....................................................... 78
Figure 4-29 The model runs at the steady-state cornering ....................................... 78
Figure 4-30 Demanded Yaw Rate Vs Front Axle No-Slip Yaw Rate ........................ 79
Figure 4-31 The vehicle lateral acceleration versus target lateral acceleration ........ 80
Figure 4-32 Yaw rate gain versus vehicle speeds .................................................... 81
Figure 5-1 Different design of in-wheel motors ......................................................... 84
Figure 5-2 Protean Electric in-wheel motor .............................................................. 85
Figure 5-3 Architecture of in-wheel motors model .................................................... 86
Figure 5-5 Westfield I-racer with in-wheel motors in ADAMS ................................... 87
V
Figure 5-6 Model definition for in-wheel motors ....................................................... 88
Figure 5-10 the model geometry of in-line motors for kinematic test ........................ 89
Figure 5-11 Dynamic check model for in-line motor ................................................. 89
Figure 5-12 A full vehicle model for drivability check ............................................... 90
Figure 5-13 Subjective results for ride testing .......................................................... 92
Figure 5-14 Eigenvalues for in wheel motors model................................................. 94
Figure 5-15 The animations for vehicle body ........................................................... 96
Figure 5-16 Front wheel hop modes for in-wheel motors ......................................... 97
Figure 5-17 The plotting of Eigenvalues scatter for in-line motors model ................. 97
Figure 5-18 Road profile for speed bump ............................................................... 100
Figure 5-19 Measured results for wheel hub acceleration ...................................... 100
Figure 5-20 Subjective results of steering behaviour ............................................. 101
Figure 5-21 Ground plane velocity during steady state cornering .......................... 102
Figure 5-22 The trajectories at cornering ............................................................... 103
Figure 5-23 Testing results for centripetal acceleration of in-line and in-wheel motors
............................................................................................................................... 104
Figure 5-24 Double Lane Change Test Course ...................................................... 105
Figure 5-25 Comparison of yaw rate for in-line and in-wheel motors in DLC ......... 106
Figure 5-26 Comparison body slip angle for in-line and in-wheel motors in DLC ... 106
Figure 5-27 Comparison of steering angle for in-line and in-wheel motors in DLC 107
Figure 6-1 Definition of torque vectoring differential ............................................... 110
Figure 6-2 Schematic of a mechanism torque-vectoring differential ....................... 111
Figure 6-3 Explanations of the left-and-right torque vectoring ................................ 112
Figure 6-4 Schematic of Driveline Control Strategy (DCS) ..................................... 115
Figure 6-5 Path curvature spline in ADAMS for cornering ...................................... 119
Figure 6-6 Double lane change test course.................Error! Bookmark not defined.
Figure 6-7 Cosine ramp lane change path visualisation ......................................... 120
Figure 6-8 Optimized lane change path with the behaviours of real driver ............. 120
Figure 6-9 Comparison of the lateral acceleration for cosine and optimised lane
change path ........................................................................................................... 121
Figure 6-10 Path curvature of lane change ............................................................ 121
Figure 6-11 Comparison of desired and actual lateral acceleration ....................... 122
Figure 6-12 Yaw Rate Gain (YRG) for three different vehicles............................... 123
Figure 6-13 Explanation of torque vectoring authority envelope ............................ 124
VI
Figure 6-14 Motor efficiency map ........................................................................... 135
Figure 6-15 Vehicle model at steady state cornering ............................................. 137
Figure 6-16 Speed compensation with driving torque ............................................ 137
Figure 6-17 Trajectory of vehicle at coasting, with driving torque and TV .............. 138
Figure 6-18 Longitudinal driving forces for rear driving wheels .............................. 139
Figure 6-19 Comparison of Yaw rate correction ..................................................... 140
Figure 6-20 Explanation of torque vectoring control at 40km/h .............................. 142
Figure 6-21 Comparisons of Yaw rate without the speed controller ....................... 142
Figure 6-22 Yaw rate with speed controlled ........................................................... 143
Figure 6-23 Body slip angle for the different driving mode ..................................... 144
Figure 6-24 Explanation of torque vectoring control at 60km/h .............................. 144
Figure 6-25 Variations of yaw rate with the speed controlled at 60 km/h ............... 145
Figure 6-26 Yaw rate variations at 60 km/h ........................................................... 145
Figure 6-27 Variations of body slip angle at 60 km/h ............................................. 146
Figure 6-28 Yaw rate variation at 100km/h............................................................. 146
Figure 6-29 Yaw rate variation with the speed controlled at 100km/h .................... 147
Figure 6-30 Comparison of body slip angles at 100km/h ....................................... 147
VII
Index for tables
Table 3-1 Front suspension hard points ................................................................. 178

Table 3-2 Rear suspension hard points.................................................................. 178
Table 3-3 A set data for tyre model .......................................................................... 50
Table 4-1 Total degrees of freedom for the model ................................................... 62
Table 4-2 Display of the eigenvalues in tabular form ............................................... 64
Table 4-3 Explanation of the natural frequencies at each mode .............................. 65
Table 5-1 Display the eigenvalues in tabular form ................................................... 93
Table 5-2 Explanation of natural frequencies at each mode .................................... 93
Table 5-3 Natural Frequencies for in-line motors model........................................... 98
Table 5-4 Comparison of natural frequencies for in-line and in-wheel motors model99
Table 6-1 The schematic of control modes ............................................................ 129
VIII
Annotations
Ax Longitudinal acceleration
Ay Lateral acceleration
b Longitudinal distance of body mass centre from front axle
c Longitudinal distance of body mass centre from rear axle
f Natural frequency
k Path curvature
Ks Spring stiffness
kw Stiffness of equivalent spring at the wheel centre
m Mass of a body
vx Longitudinal velocity
vy Lateral velocity
Fx Longitudinal tractive or braking tyre force
Fy Lateral tyre force
Fz Vertical tyre force
FD Drag force
Af Front axle slip angle
ar Rear axle slip angle
Side slip angle
Steer or toe angle
Yaw rate
Demanded yaw rate
Yaw rate error
Front axle no-slip yaw rate
T Difference in driving torque
IX
Abbreviations
ALSD Active Limited-Slip Differential

AYC Active Yaw Control
ATVC Advanced Torque Vectoring Control
AWD All Wheel Drive
ADAMS Automated Dynamic Analysis of Mechanical Systems
DOF Degrees of Freedom
EV Electric Vehicles
GRG Generalized Reduced Gradient
IC Internal Combustion
LSD Limited-Slip Differential
LQG Linear Quadratic Gaussian
ODCS Optimal Driveline Control Strategy
RWD Rear-Wheel-Drive
SC Speed Compensation
YRE Yaw Rate Error
Yaw Rate Gain YRG
X
Acknowledgement
I would like to express my gratitude to all those who gave me the possibility to
complete this thesis. I want to thank Coventry University for giving me permission to
commence this thesis in the first instance. I have furthermore to thank my second
supervisor, Dr Gary Wood, who encouraged me to go ahead with my thesis. Also I
would like to thank Mr Damian Harty for his assistance over the period of the
research study.
I am deeply indebted to my director of study Professor Mike Blundell, who gave me

huge support and helped me throughout the research project and in writing up of the
thesis.
Especially, I would like to give my special thanks to my family whose love enabled
me to complete this work.
XI
1. Introduction
1.1 Pollution and fuel consumption
Fossil oil is the most important natural resource to support the world economy.
Although recent estimates vary, there is absolutely no doubt that global concerns
about the finite nature of our oil-based energy reserves are well founded (Hirsch,
Bezdek et al. 2005). Global energy demand from all sources is expected to increase
by 1.3 percent per year on average from 2005 to 2030 (ExxonMobil 2007).
Apart from the shortage of oil supply, the concerns in growth of emissions and
pollutions have been discussed every day. In fact, the transportation sector accounts
for about 21 percent of current global fossil fuel CO2 emissions to the atmosphere-
second only to emissions from power production (IPIECA 2004). According to the
Technology Strategy Board (TSB), in the UK it is estimated that transport accounts
for 24% of the UKs carbon emissions. Road transport accounts for 80% of this figure
(IME 2009). However, the big challenge to the automotive industry is to be
responsive to both legislation-reducing emissions and the market-growing consumer
demands.
1.2 Developing electric vehicles
There has been a massive resurgence of interest in electric vehicles (EVs) over the
past decade. Many observers now see them as the long-term solution to reducing
vehicle emissions and CO2 usage in comparison to alternative approaches such as
hybrid vehicles, fuel cells or biofuels. The public perception of electric vehicles has
changed dramatically and recently announced vehicles such as the Tesla roadster
and Chevrolet Volt have reinforced the idea that they are now becoming seriously
competitive products. Not long ago, electric vehicles were still seen as niche
products and associated more with milk float technology rather than a viable
passenger transport alternative (Chan and Chau, 2001; Husain 2003; Larminie and
Lowry 2003).
1
Massive advances have occurred in battery technology although the progress has
been gradual and sustained, so that it has not commonly been perceived as a major
breakthrough. The vehicle range available with modern battery sets such as
Lithium Ion is now typically of the order of 200km, which makes electric vehicles
widely acceptable for much urban use. The high cost of the batteries is still a
problem and despite a relentless downward price trend, the battery sets are often
supplied on a leasing arrangement rather than a straightforward purchase.
As the electric vehicle market continues to grow, the chassis engineers will place
increasing emphasis on searching for dynamics control due to the new architecture
used on the electric vehicle. This process of continual improvement is central to
vehicle development of safety and handling, and has occurred for example, over
recent decades with vehicle dynamics. The industry has achieved ride comfort and
safety figures that were considered impossible twenty years ago. Of all the green
solutions, battery electric cars have the best dynamics control, individual wheel
control, of both conventional cars and hydrogen fuel-cell cars. Also the driving
efficiency is high, for example, with 1 of electricity, an EV can drive 5525 km;
while using the same amount of electricity to generate hydrogen and to drive a fuel
cell car, the distance is reduced to 1790 km (Randall, 2009).
The electric vehicle as a main test platform for this research focuses on one area of
interest in which dynamics control gains may be achievable for electric powertrains
with individual control motors. As is known, it is commonly argued that one of the
distinct advantages of an electric motor as a motive unit is its torque characteristic; it
can deliver maximum torque from zero speed and throughout the low speed range
typically up to around 2000 rev/min. Then, the available maximum torque reduces
with speed along the motors maximum power curve. This is a much better
characteristic than that associated with internal combustion engines, which cannot
deliver useful torque at low speeds and because of their relatively narrow torque and
power bands, must be used with multispeed transmissions in order to deliver tractive
power to the vehicle in a suitable form. Typical electric motors have another
desirable feature their maximum intermittent power is considerably higher than
their rated continuous power. The limiting factor is usually related to controlling the
2
amount of heat build-up. Consequently, good acceleration times can be achieved
providing they are only used for relatively short periods a situation which
fortunately is typical of normal driving.
1.3 Aim and objectives
The proposed research is focused on active yaw control for electric racing vehicles.
Due to this research being based on an industrial project from Westfield Sportcars,
the company is starting to investigate and develop a full electric racing vehicle that is
called the Westfield i-Racer. The Westfield engineers have developed the worlds
first electric race car kit that can be built at home, while also supporting the
requirement of sport racing with zero emissions vehicles.
Initially, the project vehicle with typical sub-systems such as suspension, steering
and driveline system were modelled and assembled so that the requirements of the
vehicle dynamics simulation could be fulfilled. A model audit was needed to ensure a
rigorous system was used in the following research. The audit involved calculating
the mass and inertia properties for the vehicle body and all the components in the
sub-systems, as well as finding the central of mass position for the vehicle body and
the characteristics of the spring and damper in the following section. The tyre-
sourced data is based on the Pacejka 89 version, with the manufacturers
coefficients and the tyre model being run on a computing-based tyre test rig to
indicate all characteristics of the tyre.
When building the project vehicle in computer-based software, the accuracy of the
simulation results relies on the accuracy of the model and the vehicle parameters
used to build the model. Hence, there are several methods available to validate the
vehicle model, such as kinematic and dynamic manoeuvres. In order to verify the
performance of the suspension system, a range of characteristics were determined
through simulation of a quarter vehicle model. The full vehicle model was analysed in
a number of ways that provided information to support the following investigations.
Also, the steady-state cornering manoeuvre was used to define the basic driving
characteristics of the project model.
3
A main difference in architecture for the conventional vehicles is the position of the
Internal Combustion (IC) engine, which can be located at the front, the middle or the
rear of the vehicle; thus influencing the mechanical, such as the Rear-Wheel-Drive
(RWD) or the All-Wheel-Drive (AWD). A novel architecture for the electric vehicle
driveline was designed that has the freedom to move the motors to a single location
in the vehicle, for example, by mounting the motors within individual wheels. In
addition, the electric motors can also be located in the middle of the chassis at the
front, rear or both axles. The recent conformity in electric vehicle architecture, with
high driving eff/iciency of in-wheel motors now on the market, has led to some
complacency in viewing any other architecture. Westfield Sportcars, a producer of
the in-line motors, has commissioned a series of wide-ranging studies into the
effects of vehicle performance. However, a comparison will be introduced that
includes a ride comfort check and drivability check studies using the ADAMS model.
These studies provide a comprehensive overview of the implications of in-wheel and
in-line motors in this particular racing model.
Finally, the most important part of this investigation is to develop and design a novel
Torque Vectoring control strategy. In the literature, active intelligent control systems
for achieving vehicle stability and handling have been developed and implemented to
enhance the safety and performance of the driven vehicle. Some enhancements,
such as the Active Steering System and the Electronic Stability Program, can help
the driver to retain control of their vehicles when the grip between road surface and
tyre is lost. In previous investigations, ABS based Stability Control Systems have
been the principal implementation to accomplish the safety requirements with
adverse road conditions. However, the vehicle speed is degraded while the Stability
Control System implements braking force on four wheels individually to improve the
correct position of the vehicle body. Moreover, the Torque Vectoring (TV) system
can be designed to improve the vehicle handling qualities and avoid the vehicle
speed decrease, or in other words the fun-to-drive aspect. Thus, torque vectoring
can be used to influence the driver experience.
A new control strategy, that is called Optimal Driveline Control Strategy (ODCS), will
be design and/ developed. The ODCS involves three levels of control: Desired
4
Dynamics Behaviours, Secondary Control and Advanced Torque Vectoring Control,
more details of which will be represented in the following section. Reviewing the
basic principles for TV control on the conventional driveline helps to understand how
those control strategies on the pure electric vehicle can be implemented. The
configuration of the ODCS algorithm is clearly shown and each level of control is
explained in detail.
The project vehicle model will be taken into the typical vehicle dynamic simulations,
by running the vehicle model through the steady state cornering and lane change
manoeuvres. The results will be analysed and compared against the project model
running with the different driving modes, assuming the model has the conventional
drivelines, namely Open and Limited Slip Differentials on the rear axle, so that these
will be simulated for the same manoeuvres.
5
2 Literature Review
2.1 Introduction
The previous work on Vehicle Dynamics Control, Direct Yaw Control, Torque
Vectoring and the vehicle dynamics control strategies of Electric Vehicles (EVs) are
reviewed here. The definitions and classifications for a Final Drive Unit are
summarised and the main tools for vehicle dynamics analysis are introduced. For
control strategies, the current situation and the future trends are analysed. Torque
vectoring for electric vehicles has received a substantial amount of attention and
several different designs have been proposed over the past decade. The important
dynamics control strategies for electric racing vehicles, a crucial feature in optimizing
performance and overall performance, are reviewed.
2.2 Electric Vehicle
Vehicle industries and governments around the world appear to have a high level of
interest in electric vehicles, an interest which is growing at a substantial rate. In the
past, commercial vehicles powered by the Internal Combustion engine became
available at the end of the 19th century; while at the same time, an electric vehicle
broke the world land speed record in 1899, becoming the first car to exceed one mile
per minute. At that time, there were three propulsion systems: the electric motor, the
IC engine and the steam engine. Comparing the size and complexities of these
devices, the electric motor was a clear winner; however, it used original lead acid
batteries for the energy storage, which have 300 times lower capacity than that of
the specific energy of the gasoline-driven vehicle.
Based on the global requirement to reduce emissions, both political and

technological sectors have currently been experiencing a resurgence of interest in
electric vehicles supported by emerging battery technologies. Investigations are
mainly focused on energy density, improved specific energy and rechargability
properties. The electric motor has a great advantage in torque characteristic, which
6
means the motor can provide a more desirable spread of torque over the vehicle
speed range compared to that of the IC engine, and also electric vehicle architecture
provides the shortest driveline in contrast to conventional vehicles.
Currently, the most popular design approach for electric vehicle is to connect the
motor to the driven wheels directly, with some designs requiring a transmission unit
between the motor and the wheels. In general, the characteristics of the current
electric motor have two regions intermittent peak torques and operating torques
that are comparatively lower; the high torques can provide desired acceleration from
very low speeds, and the top speed is constrained by the continuous torques.
Normally, the transmission unit with a fixed gear is used to control the top speed. In
addition, there is another area of interest in the case of an electric vehicle, which is
to investigate how to control the efficiency of the electric motor, so that particular
capacities of the motor are required, namely low speed with high torque, direct-drive
and excellent torque-power densities.
In a similar fashion, plug-in electric vehicle have also become a very topical subject.
For example, the i MiEV from Mitsubishi Motors has been commercially produced
and 200 of these vehicles have been put into the UK for test driving. Using the on-
board charger, the vehicle can be charged with a 100 V or 200 V power source in the
home. The range over one of the driving cycles, for one charge is 160 km, which is
enough for most commuting applications. For example, in the United States, half of
U.S. households have a daily mileage of less than 30 miles per day; 78% of daily
work commuters travel 40 miles or less (Babik 2006).
2.3 Drivetrain Designs
2.3.1 Conventional mechanical drivetrains
An Open differential drives the wheels to rotate at different speeds while balancing
the torques between them. A limited slip differential allows for unequal torque
distribution between the wheels, but with a fixed kinematic relationship, and the
torque can only be transferred from the wheel spinning faster to the wheel spinning
7
slower. A torque vectoring system, while retaining the ability to function as a simple
differential, incorporates a means to vary the kinematic ratio across the differential,
thus affecting the torque distribution between the wheels. Torque vector is defined as
the torque difference between the two output torques.
During cornering, the vehicle wheels rotate at different speeds. The differential is
equipped to rotate both the driven wheels with identical torque but different angular
velocity. For this device, the capability to transfer torque and rotation is through three
shafts, one input two outputs. This is found in most vehicles, which allows each of
the driven wheels to rotate at different speeds, while supplying equal torque to each
of them. When cornering, the inside wheel is rolling in a smaller circle than the
outside wheel and without the differential, the inside wheel is spinning and the
outside wheel is dragging. This scenario may cause problematic and unexpected
handling, more tyre wear, and damage to (or possible failure of) the entire drivetrain.
2.3.2 Drivetrains for electric vehicle
Novel concepts of electric vehicle layouts are gaining more and more importance.
The first generation of fully electric vehicles was based on the conversion of internal
combustion engine driven vehicles into electric vehicles, by replacing the drivetrains,
while keeping the same driveline structure; that is, one electric motor drive, which is
located centrally between the driven wheels, and a single-speed mechanical
transmission including a differential. Such a design solution is going to be gradually
substituted by novel vehicle architecture, based on the adoption of individually
controlled electric powertrains, with the unique possibility to improve the vehicle
dynamics control because of their intrinsic high and independent controllability. The
active control of electric powertrains allows the regulation of the distribution of the
driving torques in order to achieve desired steady-state and transient vehicle
dynamics characteristics. At the same time, if implemented through in-wheel motors,
these architectural solutions allow an improvement of the overall vehicle packaging
as less space is required by the powertrain.
Current electric vehicle research is investigating different powertrain configurations,

constituted by one, two, three or four electric motors with different performance in
8
terms of vehicle dynamics behaviour and energy saving targets (Novellis, Sorniotti,
Gruber,2012 and Rinderknecht, Meier,2010) . The possible architectures are shown
in Figure 2-1 (Ehsani, Gao et al.2004), from which it can be seen that mainly two
types of transmissions are used on electric vehicles: multi-gear transmission and
single-gear transmission. Currently, single gear transmissions are used on most EVs.
For example, on the Gulliver U500 design from Tecnobus, the transmission is a
single gear with a fixed ratio of 1:4.37. For configurations like Figure 2-1 (a) and (b),
an electric propulsion motor replaces the IC engine of a conventional vehicle drive
train. The multi-gear transmissions here were originally designed for an engine, not
especially for electric motors. It is perhaps surprising, but there is very little published
research on the potential benefits from connecting the motor to driven wheels
directly by using driving shafts, it can save major components in the transmission
system reduce the weight of entire vehicle and improve driving efficiency.
This item has been removed due to third party copyright. The unabridged version of the thesis
can be viewed at the Lanchester Library, Coventry University.
Figure 2-1 Possible EV configurations (Ehsani, Gao et al. 2004)
9
2.4 Vehicle dynamics control
In recent years, several researchers have invested time in, and effort on, the
improvement of vehicle dynamics control, generally focusing on agility, stability,
reliability and linearity to improve handling. These requirements have been applied to
new designs of electric and hybrid vehicles. A Limited-Slip Differential (LSD) is
based on the open differential with a type of gear arrangement that allows the
wheels to have differences in angular velocity and the driving torque. In order to
provide increased stability, alternative control systems are then considered without
the disturbance of brake-based stability control programmes. A Limited-Slip
Differential uses the electronic controller to transfer torque between the driven
wheels. Controlled torque transfer between the driven wheels can generate a yaw
moment that is able to improve the stability of vehicle; and due to the wheel torque
being redistributed without any speed reduction, this improvement in stability is less
intrusive than a brake-based stability control programme.
In general, active steering systems can help the driver to face a critical driving
situation. Zhang (2008) presented a paper that described a multi-body vehicle
dynamic model with an active steering system using a fuzzy logic control strategy.
The multi-body vehicle dynamic model was built in ADAMS and the dynamic
performance of the vehicle could be accurately predicted. The methodology of
control included active front steering and rear wheel steering by wire, to which active
steering at the front axle involved a modified steering angle added to the driver input,
and the rear wheel steering was controlled by wire. This combination was effectively
used to control both body-slip angle and the yaw rate. The controllers used in the
active front and rear steering control were based on a set of fuzzy logic rules to
adjust the body-slip angle and the yaw rate. Optimization of the fuzzy logic control in
both the active front and rear steering system was also represented. Thus, the
simulation results indicated that active front and rear steering using a fuzzy control
logic strategy enabled improved handling and stability of the vehicle comparing the
four-wheel steering with front wheel steering only.
Modification of the vehicle dynamics can also be achieved by controlling the

distribution of lateral forces using the combinations of front and/or rear steering
10
angles in four-wheel steer by wire (Ackermann and Sienel, 1993, Ackermann et al.,
1995, Kohen and Ecrick, 2004 and Vilaplana et al., 2005). Moreover, the
improvement of the vehicle safety capability for emergency avoidance, using the
optimization of both the direct yaw control and active steering control is another
research area of interest (Mokhiamar and Abe, 2002).
According to these previous works, the control designs are only for the development
of the yaw stability control algorithm. The control is based on a model with the
desired vehicle response. The model is a basic vehicle model that can be used to
calculate the desired yaw rate based on the steering input, vehicle speed and road
surface. Moreover, the Limited-Slip Differential designs generate a required torque to
transfer across the axle depending on the error between the desired and actual yaw
rate of the vehicle. The error is fed through a feedback controller, but the Limited-Slip
Differential only develops the yaw moment in the under-steer condition. Also Four-
Wheel Steer and Active steering systems have same limitation.
2.4.1 Direct Yaw Control
Due to unusual external conditions there can be unexpected dangerous behaviours

in vehicle yaw dynamics, such as unexpected side-wind force, different road surface
texture on left-right wheels, and emergency avoidance. Moreover, under-steer may
degrade the handling performance in cornering manoeuvres and cause discomfort to
the human driver. There are, however, a few solutions available in recent years to
solve these issues, and the task is still to carry out intensive research activities in
both practical and analytical studies (see e.g. Brner and Isermann, 2006, Colombo,
2005, Gaspar et al., 2005). In this case the purpose of the study was to modify the
vehicle dynamics and exploit the best combinations in longitudinal and lateral tyre
forces. Also, using the uneven longitudinal driving force on the left and right sides
can control the yaw rate. This approach involved different technologies, for example,
Anti-Lock Braking System (ABS) and Electronic Stability Program (ESP) (Zanten,
2000 and Zanten, 1995) or torque vectoring control using active differentials
(Assadian and Hancock, 2005, Colombo, 2005 and Gerhard, 2005).
11
Assadian, Hancock and Best (2010), describe some developments on mechanical
limited slip differentials which provide a low cost traction solution. However, their
passive nature means that mechanical limited slip differentials cannot adapt to
different conditions and their yaw moment generation potential cannot be used for
vehicle handling or stability control. Active limited slip differentials are becoming
popular as they are able to exploit this potential and also achieve a better traction
compromise due to their ability to adapt to different scenarios. The development of a
control algorithm for an Active Limited-Slip Differential (ALSD) fitted to a RWD sports
saloon vehicle. The ALSD uses a wet friction clutch unit to transfer the torque across
the driven axle, and a driven actuation system with an electric motor controls the
clamping force on the clutch unit through a ball and ramp device.
However, the Limited-Slip Differentials use the wet clutch unit to provide a controlled
left-and-right torque distribution on the front or rear axle, and four-wheel torque
distribution, thus resulting in improved traction control and yaw stability control
performance without being intrusive for the driver. The case speed is equal to
and the clutch always transfers the torque from its faster to its
slower shaft. The direction of torque transfer is determined by the difference in the
wheel speed across the axle. It is restricted to over-steer compensation only when
being used as a yaw stability control device, since only an under-steer torque can be
generated.
Active Yaw Control (AYC) and Super-Active Yaw Control developed by Mitsubishi
Motors (Ushiroda,2003) based on an active differential to modify the torques at the
driven wheels are reviewed in this section. These products were designed and
implemented in the series of the Mitsubishi Lancer Evolution cars, which used a
planetary gear-set, also being used in several Mitsubishi concept models, to support
a greater torque distribution than that of the existing systems.
The simulation results, as shown in Figure 2-2, show that the maximum cornering
performance is obtained by using an AYC model without limitations on torque
transfer. This concept is based on a left-and-right torque vectoring system and the
driving torque is optimally controlled depending on the vehicle conditions. A
comparison of the vehicle with and without the AYC is presented in this paper; and
12
each of the curves shows the maximum lateral acceleration during the cornering
manoeuvre for a given acceleration in the longitudinal direction. It is clear to see that
the acceleration region of the optimally controlled AYC vehicle is 25% larger than
that of the vehicle with only AYC. The torque transfer for the optimally controlled
AYC is 1.8 times larger than the torque transfer amount with AYC alone. Thus, the
increment in the amount of the torque transfer offers a target for which the authors in
this research have identified potential methods to achieve.
This item has been removed due to third party copyright. The
unabridged version of the thesis can be viewed at the Lanchester
Library, Coventry University.
Figure 0-2 Acceleration cornering performance (USHIRODA 2003)
Three different torque vectoring strategies which summarise the strategies explained
in above the research have been implemented: i) constant torque distribution
(referred to as the baseline vehicle); ii) torque proportional to the wheel vertical load;
iii) torque distribution which allows achieving the same longitudinal slip ratio at each
wheel.
2.4.2 Torque Vectoring System
Torque vectoring control can have a major impact on the general driving experience.
Most of the time the driver operates the vehicle in steady-state, or slowly varying
conditions, at lateral acceleration levels as below 0.5 g (Pacejka, 2006). During
these sub-limit conditions, the continuous yaw moment control can significantly
improve the vehicle cornering response. As recently pointed out in (Crolla, 2012),
13
despite the significant volume of theoretical studies of torque-vectoring on vehicle
handling control, there is no widely accepted design methodology of how to exploit it
to improve vehicle handling and stability significantly. To address this issue, novel
tools for the design of torque vectoring control systems have to be proposed and
assessed.
Vehicle steady-state cornering response is usually assessed in terms of its

understeer characteristic, where the dynamic steering wheel angle is the difference
between the actual steering wheel angle and the kinematic steering wheel angle
(Gillespie, 1992). In general, in a passenger car, the dynamic steering wheel angle
increases monotonically and nearly linearly up to a value of lateral acceleration of
about 0.5 g for high friction conditions. Correspondingly, the understeer gradient of
the vehicle is nearly constant. Beyond this linear region is non-linear and tends to an
asymptotic value corresponding to maximum lateral acceleration when the tire friction
limits are reached. In contrast to vehicles without torque vectoring control, where the
specific understeer characteristics are determined by the tyre properties, geometrical
and inertial parameters and the suspension elasto-kinematics (Reimpell, 2001 and
Milliken, 2002), the understeer characteristics of a vehicle equipped with a TV
system can be designed to achieve almost any desired response. For example, the
understeer gradient in the linear part of the characteristic could be imposed. Also,
the width of the linear region could be increased, or the maximum lateral
acceleration could be altered, with the constraints dictated by tyre friction limits
(Zorzutti, 2007).
In addition to the advantages during pure cornering manoeuvres, continuous TV

control has the potential to improve the handling response of a vehicle while braking
or accelerating. Despite the significant influence of accelerating and braking, the
understeer characteristics for non-zero longitudinal acceleration are normally not
considered and analysed. This restriction mainly results from limitations imposed by
the typical vehicle dynamics simulation techniques or testing procedures used to
derive the zero longitudinal acceleration cornering response plots, namely, skid-pad
tests or ramp-steer manoeuvres.
14
Yaw control is effective in order to realise the active safety philosophy that makes
the likely occurrence of an accident small; and the yaw control technology by using
brake based systems has been developed in a large number of products. As a next
step, direct yaw control can be achieved using the right-and-left torque vectoring
control. This system can directly control the yaw moment acting on a vehicle by
vectoring the torques between wheels on either side with minimum energy loss.
Therefore, the strong point of this system is to be able to improve the stability of the
vehicle from the normal condition to the high marginal condition seamlessly
(Ikushima,1995).
This item has been removed due to third party copyright. The unabridged version of the thesis can
be viewed at the Lanchester Library, Coventry University.
Figure 0-1 Vectoring torque acting on rear right and left wheels (Sawase and Ushiroda, 2007)
A concept is shown in Figure 2-3; the torque flow between the right and left wheels
can be controlled by a device, so that it can make on one side of the driving forces
small, and the other side driving forces large. The difference in the longitudinal
driving torque generated on the right-and-left wheels can control the yaw moment
that acts on the vehicle, even if the engine torque and/or braking force are applied.
This item has been removed due to third party copyright. The unabridged version of the thesis can be
viewed at the Lanchester Library, Coventry University.
//Figure 0-2 Effect of torque vectoring (Sawase and Ushiroda, 2007)
15
Figure 2-4 shows the relation between the tyre maximum friction force, the driving
force, and the maximum cornering force of the right and left wheels during cornering.
When the vehicle is cornering to the left, lateral acceleration causes the left tyre
maximum friction force (shown as the radius of the tyre maximum friction circle) to
decrease and the right tyre maximum friction force to increase. Therefore is
smaller than . The left tyre maximum friction force is the same and the left
driving force D assumes an equal state in the case of a vehicle without torque
vectoring, as shown in Figure 2-4 (A). In this state, the right wheel can only generate
the maximum cornering force in (A), because the right tyre maximum friction
force is bigger than the right driving force . The left driving force is - and
the right driving force is , so that the left wheel can generate the
maximum cornering force, and the right wheel can generate the maximum
cornering force . The right and left wheels total maximum cornering force
difference between a vehicle with torque vectoring and one without torque vectoring
by , is expressed by the following equation,
(2.1)
In this equation, when is increasing from zero, becomes maximum value at
(2.2)
Thus, it is shown that a torque vectoring increases the total maximum cornering
force.
The calculated influence of the right-and-left torque vectoring control in the different
types of driveline, namely, the front wheels only, the rear wheels only, and both front
and rear wheels in FWD, RWD, and AWD vehicles is represented. The effect is
evaluated by calculating the vehicle dynamics limit, the maximum acceleration and
the cornering ability. The right-and-left vectoring torque, which is needed to increase
the vehicle dynamics limit, is also calculated. Also, the application to the front wheels
is more effective for FWD vehicles. On the other hand, the application to the rear
wheels is more effective for RWD and AWD vehicles.
16
Figure 0-3 Left-Right Torque Vectoring Concept
Figure 2-5 shows the concept of left-and-right torque vectoring control in which this
approach has the braking force applied to the left wheel and the same magnitude of
the driving force applied on the right wheel. Thus, it is able to control the yaw rate
directly as required at any time, and the control is without any constraints from the
level of the engine torque, and any conflicts between the torque vectoring control
and the operation of the driver.
This item has been removed due to third party copyright. The unabridged version of the thesis can be viewed at the
Lanchester Library, Coventry University.
Figure 0-4 Schematics of recent torque vectoring systems (Wheals, 2004)
Figure 2-6 shows a comparison between the Ricardo Torque Vectoring systems and
alternatives such as the Mitsubishi EVO VIII device and the Mimura device:
All designs provide permanent drive to the wheels via a differential when the
actuation system is inactive.
17
The Ricardo design uses two brakes whereas the Mitsubishi and Mimura
designs both use two clutches. It describes a single brake design.
The Mitsubishi and Mimura designs both use joined planet gears within the
geared stage that force a speed difference between the outputs.
The Ricardo Torque Vectoring device uses joined sun gear, which requires
the use of additional annulus gears which are not required by the other
designs.
2.5 Handling Control for electric vehicles
Electric vehicles can have different topological layouts with in-wheel or on-line motor
drives. This design flexibility, combined with the possibility of continuous modulation
of the electric motor torque, allows the implementation of advanced torque-vectoring
(TV) control systems. In particular, based on the individual wheel torque control,
novel TV strategies aimed at enhancing active safety (Kang, 2011, Nam, 2012 and
Jonasson, 2011) and fun-to-drive qualities (Gruber, 2013) in all possible driving
conditions can be developed. Indeed, by directly controlling the yaw moment through
the actuation of electric drivetrains, a TV system extends the safe driving conditions
to greater vehicle velocities during emergency transient manoeuvres than a
conventional vehicle dynamics control system based on the actuation of the friction
brakes (Tseng, 1999, Doumiati, 2011). Different electric vehicle layouts are currently
analysed for the demonstration of TV control strategies, including multiple
individually controllable drivetrains (Xiong, 2009, Wang, 2009, Akaho, 2010,
Tabbache,2011, Chen,2013) or one electric motor per axle coupled with an open
mechanical differential or a TV mechanical differential.
Torque Vectoring control structures are usually organized according to a hierarchical

approach as shown in Figure 2-7. A high-level vehicle dynamics controller generates
a reference vehicle yaw rate, which is adopted by a feedback controller in order to
compute the reference tractive or braking torque and yaw moment. The feedback
controller is either based on sliding mode (Canale, 2005, Ferrara, 2009), linear
quadratic regulation (Zanten, 2000), model predictive control (Chang, 2007) or
robust control (Yin, 2007) . A feedforward contribution, , for example based on
maps, can be also included, as shown in Figure 2-7, in such a way that the control
18
yaw moment is given by , where the feedback term
compensates the inaccuracies, the disturbances or the variation of the vehicle
parameters (such as vehicle mass, position of the center of gravity, etc)
considered for the derivation of the feedforward maps.
Figure 2-7 Functional schematic of a typical TV controller for a FEV with multiple
individually controllable drivetrains also illustrated in (Xiong, 2009 and W ang,2009).
At a lower level, the objective of the control allocation is to generate appropriate

commands for the actuators in order to produce the desired control action in terms of
traction or braking torque and yaw moment. When the number of actuators is larger
than the number of reference control actions, the control allocation problem can be
solved by minimizing an assigned objective function. This is achieved with simplified
formulas based on the vertical load distribution (Tanaka, 1992 and Mutoh, 2012) or
with more advanced techniques such as weighted pseudo-inverse control allocation
(Tabbache, 2011 and Yim,2012), linear matrix inequality (Fallah, 2013) or quadratic
programming with inequality constraints (Tjonnas, 2010). The optimization
algorithms most commonly employed for on-line control allocation schemes are
active set, fixed point and accelerated fixed point. The published methods are shown
to be successful, but their application and analysis are limited as their tuning is
carried out through the optimization of the vehicle performance during specific
maneuvers (Naraghj, 2010) and not the full range of possible operating conditions.
More importantly, the effect of the possible alternative formulations of the objective
functions for control allocation on the overall performance is not explored in the
literature.
For example, Kim (2007) designed a new control algorithm for the stability
enhancement in which an electric vehicle with four-wheel-drive used the rear in-
wheel motor driving, regenerative braking control, and electrohydraulic brake (EHB)
19
control. The control algorithm is based on a fuzzy-rule-based control that can
minimize the errors of the body-slip angle and the yaw rate. A co-simulation of
ADAMS and Simulink was used in this research, in which the vehicle was modelled
in ADAMS with the suspension system, tyres, and steering system to describe the
dynamic behaviour of the vehicle. Moreover, the driveline components of the given
vehicle with the control algorithm, such as the motor, engine, transmission and
battery, were modelled in MATLAB Simulink, and again only the chassis elements
modelled in ADAMS. The simulation results showed that the combination of the rear
motor driving and regenerative braking can improve the stability performance and
the driving efficiency of the vehicle.
Also, Shino and Nagai (2001) have investigated the use of direct yaw rate control
using a brake-based system to distribute the driving torque, thus improving the
vehicle dynamics of electric vehicles. Based on their research, the design used the
architecture of the electric vehicle with in-wheel motors that can implement the
control strategy to the vehicle to fulfil the requirements of the control performance.
Fundamentally, the control strategy based on a model following controller is able to
impel the vehicle to follow the desired dynamics. The control strategy included a
feed-forward body-slip angle regulator and the feedback control for the yaw rate. The
vehicle with the new control strategy has been simulated in several computer-based
manoeuvres to test the performance of the control. The validations clearly showed
the dynamic behaviours of the electric vehicle have been enhanced, particularly in
the handling and stability; also the improvement can be seen when the vehicle has
been put through the different conditions of the road surface.
Pinto, Aldworth and Watkinson (2011) carried out research to develop a yaw motion
control system based on torque vectoring with twin rear electric motors, with the
main objective of enhancing the driving dynamics of a hybrid vehicle without
compromising requirements on low emissions, safety or driver feedback. The distinct
advantages of the system are investigated with simulation tools and verified with field
measurements on MIRAs prototype Hybrid 4-Wheel Drive Vehicle.
20
This item has been removed due to third party copyright. The unabridged version of the thesis can be viewed at
the Lanchester Library, Coventry University.
Figure 0-5 W ide lane change at 70km/h (L Pinto, Aldworth and W atkinson , 2004)
The result of the testing, which measured the yaw rate, hand-wheel angle, side-slip
velocity, lateral acceleration and estimated side-slip angle are measured in a typical
evasive manoeuvre, is shown in Figure 2-7. Results show effective under-steer
compensation, enhanced agility, increased cornering speed, improved yaw damping,
and a possibility to negotiate tight corners with drifts controlled by the drivers
steering input. The high yaw authority is compensating under-steer, the possibility of
enforced optimal yaw tracking in sub-limit driving, and the high potential for ease of
integration with an existing ESC system. It makes the system suitable not only for
sport applications but also for enhancing the everyday driving manoeuvrability of
standard compact and subcompact vehicles. In its simplest version, it can be
retrofitted to a standard FWD vehicle, together with a relatively small battery, and it
can also be used to provide drivability functions such as launch support and 4WD
mode, or it can be fully integrated into a proper hybrid power-train management
system.
21
2.5.1 Driver Model
In many situations it can be beneficial to do analysis of the driver using a virtual

representation. For real vehicle tests you can replace the driver with a steering robot,
this being superior to the human in precision and repeatability for pre-defined control
of the vehicle. If the driver is replaced with computer models, this allows the
processing of large batches of tests in desktop computer simulation programs. For
the driver input to the vehicle model you can use either open loop pre-defined
steering or driver models, which more accurately represent the human driver and
their limitations.
One of the first recognised model based driver descriptions is to be found in an early
article (Gibson and Crooks, 1938). McRuer is one author who has had great
influence on control-theory-based pilot models and driver model development, e.g. in
(Westbrook,1959, McRuer and Wier 1967, and McRuer, 1980). Other authors
(Fiala,1966, Mitschke, 1972, and Allen,1987) who have pioneered the development
of driver models. In the early eighties, MacAdam presented his work on optimal
control, which provided a much appreciated method for predicting vehicle movement
which allowed good path following. Sharp and Casanova (2000) have among other
things contributed with mathematical model and optimal control model development
(Sharp and Valtertsiotis, 2001). Another driver modelling approach was given by
Cole, who has studied neuro-muscular activities in the drivers steer control and
implemented this research in driver models (Pick and Cole, 2003).
Driver models have been utilised in a number of different applications in the

automotive field, such as safety, handling and fuel consumption. Driving safety is an
important area of interest since the inception of the first vehicles, and driver models
are now being used to improve safety. An understanding of driver behaviour when
alert is needed, so that deviation from this type of behaviour may indicate that the
driver is performing in an impaired state. As an example of this the frequency of
steering correction (Paul) can be used as an indicator of fatigue. Similar driver model
applications related to safety include predicting when an unsafe driver state may
occur due to tiredness, distraction or impairment. In (Onken) a driver model is used
to compare predictions with data from a computer vision system to decide whether
22
there is adequate distance from the car ahead and provide a warning if it is deemed
to be a dangerous situation.
When simulating vehicle performance over a drive cycle, it is important to consider

the effect of driver behaviour on the fuel consumption and emissions. The
regulations of the New European Driving Cycle (NEDC) stipulate that the velocity
profile is followed within a tolerance band to allow for the reaction time and sensitivity
of the driver. Compared to a closed loop PID controller historically used in simulation
to closely follow the velocity profile these driver deviations will have an effect on fuel
consumption and emissions (Froberg, 2008 and McGordon, 2011). These deviations
occur despite the use of professional test drivers with great experience of the cycle
to be followed. The research into the motivations for particular driving behaviour can
be extended to categorise certain types of behaviour into a driving style and analyse
how particular styles affect traffic flow (Treiber, 2013), accident rates (Lajunen, 1997),
and fuel consumption and emissions (Holmn,1998). Average acceleration and
standard deviation of acceleration are used in (Langari, 2005) to identify driving style,
categorised as calm, normal or aggressive using a fuzzy logic based system.
Similarly the derivative of acceleration, known as jerk, is used in (Murphey, 2009) to
classify driving style as calm, normal or aggressive.
In order to model driver performance and to determine if a driver's behaviour can be

considered appropriate, any deviation from desired behaviour can be viewed as an
error. Driving errors have some relation to vehicle safety. Driving errors can occur at
all 3 levels of the driver behaviour and are classified as either being slips/lapses or
mistakes (Parker, 2007). At the knowledge-based performance level of driver
behaviour errors are considered to be mistakes, these errors occur due to incorrect
or limited knowledge of the driving situation which results in the wrong course of
action taken by the driver. At the rule-based level errors are also classified as
mistakes, generally these errors are the result of misapplying a certain rule to the
given situation. Finally at the skill-based level, errors are regarded as slips or lapses.
In order to quantify any errors certain measures are required. Several methods for
time related measures are discussed by Horst (2007) who differentiates between
methods that can be used for either lateral control. For lateral control of the vehicle
the most heavily discussed measure is the Time-to-Line Crossing which, as the
23
name suggests, gives the amount of time before a vehicle crosses the line marking a
lane and wanders over into another lane. To calculate the Time-to-Line Crossing, the
lateral position, heading angle and speed are used. The driver has control over these
parameters through the steering angle.
2.5.2 Importance of control strategy
The design of control strategies for active or semi-active differentials is needed to

improve the handling performances of a vehicle (Cheli, Giaramita, Pedrinelli, 2005,
Resta, Teuschl, Zanchetta, Zorzutti, 2005, Cheli, Pedrinelli, Resta, Travaglio, 2006).
The purpose of this section is to discuss recent investigations (since 2000) and
conclude the current control strategy for the torque vectoring system. The discussion
will focus on knowledge of the control and the effectiveness for both situations of
simulation and practice.
Figure 0-6 Control strategy for brake-based torque vectoring (Sabbioni, Kakalis and
Cheli, 2010)
Some researchers (Sabbioni, Kakalis and Cheli, 2010) have made an enhancement
in the performances of present control systems, which is by adding a Cybe Tyre
into a Brake Torque Vectoring control strategy. The function of brake-based torque
vectoring is to apply differential braking to the vehicle driven wheels in order to
generate a yaw moment and in a simultaneous adjustment of the throttle valve to
24
avoid undesired speed reductions. The research has been developed by using
validated numerical models and carrying out a series of test-runs. The additional
information is able to exploit the potentialities of a control strategy oriented at
enhancing the vehicle handling performance, and it can ensure control action
robustness and vehicle safety under several road conditions such as dry, wet, snow,
ice. In Figure 2-8, depending on the manoeuvres at steady-state and transient, the
control logic identifies the braking torque and the eventual adjustment of the throttle
valve needed in order to improve vehicle handling.
Kaiser, Holzmann, Chretien and Korte (2011) designed a vehicle model with 14
degrees of freedom, and the two individual electric motors mounted at each wheel
centre can apply the positive and negative torque. This means it is possible for the
driven wheels to be accelerated and braked independently. As is known, the driving
torque from the electric motors has an extreme high level of quick response and
accuracy for actuating the driving wheels (Milehins, Cheng, Chu, and Jones, 2010).
In their paper, a control strategy for hybrid electric vehicle with torque vectoring
control has been presented.
In Kangs paper, an optimal torque vectoring strategy for 4WD electric vehicles (EV)
has been described in order to enhance vehicle manoeuvrability and lateral stability
with the vehicle rollover prevention (Kang, Yi and Heo, 2012). They designed the
4WD EV driving with an in-line motor at a front driving shaft and in-wheel motors at
the rear wheels, and the driving control algorithm involves three parts: a supervisory
controller, an upper-level controller and an optimal torque vectoring algorithm. At the
first level of control, the determinations such as the control functions, operation
region of control and desired behaviours can be made, and the upper-level controller
calculates the magnitude of driving force and yaw rate to follow the desired
behaviours. The optimal torque vectoring algorithm determines actuator commands.
The optimal torque vectoring algorithm is developed to map the desired driving force
and the yaw rate to the actuators, taking into account the actuator constraints. Also,
a wheel slip controller is designed to keep the slip ratio at each wheel below a limit
value.
25
This item has been removed due to third party copyright. The unabridged version of the thesis can be viewed at the
Figure 0-7 Schematic diagram of the driving control algorithm (Juyong Kang, Yi
k yongsu and Hyundong Heo, 2012)
In order to describe the overall control architecture in their research simply, Figure 2-
10 proposes a schematic diagram of the driving control algorithm, in which a control
algorithm is designed to improve vehicle manoeuvrability and lateral stability. The
control architecture involves three control levels, namely the supervisory controller,
the upper-level controller and the optimal torque vectoring algorithm. The first level of
control is to determine the control mode and desired behaviours, and the upper-level
controller is able to minimize yaw rate error and modify the driving force. As follows,
the inputs from upper-level control and wheel slip control are applied to the optimal
torque vectoring algorithm.
2.6 Concluding remarks
Interest in the Electric Vehicle has increased rapidly over recent years, from both
industrial and academic viewpoints. Research and development efforts have been
focused on developing new concepts and low cost systems, but this has proved
difficult primarily because of high battery costs. The current design trends in the
improvement of the vehicle dynamics have been summarised. Historically, the role of
the active yaw control in the overall development of vehicle dynamics control
technology has commonly been the main area of focus. Active steering control and
braking-based stability control have actually played the main crucial roles in
26
controlling the yaw moment of vehicles and ensuring good safety, driveability and
manoeuvrability. It is concluded that it has been an area of rapidly changing
technology over the past decade. But perhaps more importantly, this is certain to
continue over the next decade.
The overall conclusion about the current mechanical drivelines is that whilst the first
generation, open differential has proved to be adequate in the earlier passenger car
sector, the limited-slip differential is set to have the ability to transfer the driving
torque from the faster wheel to the slower wheel. Focusing on left-and-right torque
vectoring systems, several types of torque vectoring differentials were reviewed. For
the electric vehicle, the effect of direct yaw control on the vehicle performance and
driving safety has not yet received much attention in the current research.
As mentioned previously, there are many different control strategies that have been
used to solve the dynamics issues of the electric vehicle, but the overall designs can
be summarised in only three generic types that are likely to have a future in the short
to medium term:
i) Close-Loop Control still the most important method being used in

practical and prototype control systems.
ii) Multiple Controllers useful applications which enable the enhancement
of the performance.
iii) Optimization to define the optimal performance and integrate into the
close-loop control system design.
Almost all this recent work reviewed here has used standard vehicles and only
investigated understeer characteristics as a basis for comparisons. Whilst there are
good reasons to justify this, it nevertheless raises a driving response concern of
whether the industry is designing cars around arbitrarily selected conventional
Torque Vectoring control strategies using a full electric vehicle rather than around
architectures of electric vehicle, Linearity and efficient control strategies demands.
There remains further scope for research into handling issues in the lateral
acceleration limitation and direct yaw control associated within racing conditions. A
significant goal of benefit to the industry would be a better understanding of the
27
subjective/objective correlation of handling. There are three steps used to implement
the torque vectoring control: defining reference model, close-loop control, and
optimising the output. In this research, the close-loop control with PID controller and
Generalized Reduced Gradient (GRG) algorithm are chosen as the methodologies
most suitable for further investigation and implementation.
28
3. Modelling Electric Vehicle in ADAMS
3.1 Introduction
Vehicle design and development relies greatly on computer simulation to study

general trends before investing heavily in actual experimental testing. Testing
typically requires multiple vehicles, numerous sets of tyres and expensive
instrumentation to confirm the performance of a new design. After all this, there is
still a possibility that the results will show that the desired performance cannot be
obtained, leading to further expense. To alleviate the expenses associated, a model
is first created and simulations are run, to make sure that the design is foundational
and that one may indeed go ahead and invest in the experimentation.
In this Chapter, the project vehicle with typical sub-systems such as suspensions,
steering and driveline systems will be modelled and assembled so that the
requirements of the vehicle dynamics simulation can be fulfilled. A model audit is
needed to ensure a rigorous system model can be used in the following research.
The audit involves calculating the mass and inertia properties for the vehicle body
and all the components from the sub-systems, as well as finding the centre of mass
position of the vehicle body and the characteristics of the springs and dampers. The
tyre-sourced data is based on the Pacejka 89 version of the Magic-Formula tyre
model, with the manufacturers coefficients and the tyre model being run on a
computer-based tyre test rig to indicate all characteristics of the tyre.
3.2 Project Vehicle
Due to the research being based on an industrial project from Westfield Sportcars,
the company is starting to investigate and develop a full electric racing vehicle that is
called the Westfield i-Racer (shown in Figure 3-1). The Westfield engineers have
developed the worlds first electric race car kit that can be built at home, while also
supporting the requirement of a sports or racing vehicle with zero emissions.
29
The main specifications of the Westfield i-Racer are reasonably similar to those of
typical wingless sprint cars. The Westfield i-Racer is short 3.6 metres in length and
1.635 metres in width but not slim as midget racing cars. Moreover, the Westfield i-
Racer is heavier than a Formula One car with a limited weight of 770 kg, because of
the 200 kg of lithium iron phosphate batteries that have to be carried with the vehicle.
Figure 3-1 Architecture of Westfield i-Racer
3.2.1 Driveline
The Westfield i-Racer, equipped with two YASA-750 electric motors, has a neoteric
architecture of which the two in-line motors are located on the rear axle connected to
each of the rear wheels, as shown in Figure 3-2. The YASA motors are specially
designed for sports car and racing applications, thus the low speed with high torque,
direct-drive and excellent torque-power densities are the particular capacities of the
motors and the motors has been used in several applications and industries.
Specifically, the motor has the high torque (Peak torque 800Nm/ Continuous 400Nm)
30
and high power (Peak power at 700v is 200kw and continuous 75kW) with a range of
revolution speed 0-4000rpm. These characteristics make it very suitable for diesel
generation and direct-drive applications. However, there are some limitations of the
project vehicle, such as the top speed being constrained to 115 mph and also each
motor being limited to 45kw.
Driving Motors
Location
Figure 3-2 The position of the electric motors
An innovative modular design of the batteries has an energy capacity of 23 kWh,

including up to eleven sealed Lithium Ion Phosphate units (48V), which contribute to
the handling of the project vehicle, and which also can meet the stringent safety
standards, as shown in Figure 3-3. Particularly, the operational range of the project
vehicle is expected to be 50-60 miles so that the racing time is about 25 minutes.
Battery
package
Figure 3-3 The position of the battery package
31
3.2.2 Chassis and Suspension system
Westfield Sportcars have completed a light-weighting project working with tubing

specialist Reynolds Technology, creating a new car chassis and wishbone by
replacing much of the traditional mild steel with alloy tubing as used in Tour de
France race bicycles, as shown in Figure 3-4.
Figure 3-4 The pictures for the chassis and suspensions
The design team chose to reduce the weight of the chassis and the wishbones by
examining the possibility of using new thin-walled alloy tubing developed by
Reynolds in place of the mild steel tubes used previously, as shown in Figure 3-5.
a) Front b) Rear
Figure 3-5 The pictures of the front and rear suspensions
32
3.3 Modelling project vehicle
In this section, the main systems will be modelled and assembled in computing-
based software ADAMS that can allow a number of vehicle subsystems to be
modelled and simulated, namely the driveline, chassis and vehicle body. The
modelling of a road as an element is also included to constitute the entire vehicle
model; and modelling the suspension system is the most important part in this
process.
The overall vehicle model is prepared using the Automated Dynamic Analysis of
Mechanical Systems (ADAMS) tool from MSC.Software. ADAMS is a widely used
tool for computing the large amplitude non-linear dynamic behaviour of systems such
as ground vehicles. For this exercise, the ADAMS/Car environment was not used in
order to make the modelling practices more transparent the extensive use of
templates and sealed up run-time procedures can sometimes make it difficult to
understand the exact performance of the model when using the Car environment.
The process which has been used for modelling is planned to easily make some
modifications later in the design. So, using the graphical user interface of
ADAMS/View to create the full project vehicle as shown in Figure 3-6, the starting
point is to create hard points to indicate the different key locations of the subsystems.
As follows, these hard points are used to create the linkages; then the different joints
are added to connect the components together to finalize the system. Finally, all the
components of the project model have the mass and inertia properties.
Figure 3-6 Westfield i-Racer model in ADAMS/View
33
In the following section, a model audit is carried out so that it can ensure that the
model is more reliable and precise. More discussions are involved in this section and
the auditing aspects of the model are listed as below:
Sprung mass (The vehicle body)

Centre of Mass Position (The vehicle body)
Inertial Properties (All the components)
Un-sprung Mass
Spring and Damper Characteristics
Modal Solution
Total Degrees of freedom
Eigen Solution
3.3.1 Vehicle Body
The first concern for describing the full vehicle is to define the data set of the vehicle
body, namely the centre of mass position, the inertia properties and the masses, to
fulfil the requirements of the vehicle dynamics simulation. In this particular vehicle
model, the mass data of the vehicle body includes the mass of the body frame as
well as the additional masses of the Front Battery Module, Motors, Bracket, Coolant
and Controllers, and a passenger may even be added. Based on the project vehicle
with a conventional architecture, a summary of the main data collections will be
exhibited, namely that of the suspension geometry and the structure of the vehicle
body. The original geometric data was captured using a single point Faro arm
measuring device in the Coventry University Workshop, and all the dimensions of all
the systems such as the suspensions, the springs and dampers, and the body were
then processed using ADAMS to collate all of the data and achieve the single point
data, as required for the computer simulations.
The mass of the Westfield i-Racer model was measured in the Coventry University
Laboratory. It is important to note that the battery pack module, the complete rear
motor controller unit and the vehicle body panels, as additional masses to the mass
of the full vehicle, were assessed. The front, rear and full vertical loads are
measured so that the mass and the centre of mass position in the longitudinal are be
34
defined. Here is a method to attain the centre of mass height, which is to lift the rear
axle of the vehicle so the vehicle centreline from front to rear creates a certain angle.
A diagram of this is shown in Figure 3-7; reproduced with permission from Milliken.
This item has been removed due to third party copyright. The unabridged version of the
thesis can be viewed at the Lanchester Library, Coventry University.
Figure 3-7 The method for locating vehicle centre of mass height (Milliken, 1955)
The solution for vehicle height is given by equation (3.1).
( )
(3.1)
35
RL is the radius of the front tyre, W is the total weight of the vehicle, and W F is the
weight of the front of the vehicle during the test. The above method gave the centre
of mass height at 232 mm above the road surface. Using this approach, where the
total masses of the vehicle were the summation of the major components, the mass
centre position can also be found by using the method described above.
The inertia properties of the vehicle body need to be determined because the
dynamics of an actual vehicle are significantly affected by the yaw moment of
inertia . This data generated mainly focused on two main areas in the mass of the
chassis and four wheels. There is a detailed model to represent the mass and inertia
properties for the body that involves the overall masses of an approximate chassis,
battery Module and a motor controller which is provided by Westfield Sportcars, as
shown in Figure 3-3 above.
3.3.2 Suspension System
A completed suspension system as a most important system of the full vehicle model
will be established. In order to achieve a suspension model representing the one
installed on the actual vehicle, the use of powerful multi-body systems analysis
programs can be used. In the following discussion, modelling of a double wishbone
suspension system at both the front and rear of the vehicle is referred to as a
Linkage Model. Thus, the components of the suspension as rigid bodies can be
modelled in detail and the actual front and rear suspensions as shown in Figure 3-8.
Figure 3-8 The actual front and rear suspensions
36
The SAE coordinate system is used while modelling the vehicle, with x backward, y
to the right, and z downward. The origin is located directly below the centre of the
vehicle body geometry. SI units are used, i.e. metres for length, Newtons for force,
kilograms for mass and seconds for time. The data for the hard point locations are
measured by using a single point Faro arm-measuring device in the Coventry
University Laboratory. Table A5.1 lists the hard point coordinates of the right front
side suspensions hard points. The left side hard points are coded as mirror images
of the right points on the x-z plane. Hard-coding the left side as a mirror image of the
right makes it simpler to make changes in the suspension geometry, as changing the
coordinates of a point on the left side automatically maintains the symmetry of the
right side. Table A5.2 lists the hard point coordinates for the rear left suspension.
Here again the left side is hard coded as the mirror image of the left. The tables can
be found in Appendix 5.
The linkages of the suspension are generated by using the various shapes of the
rigid body in the Tool-box. The hard points were defined as being used to locate the
positions of each suspension linkage and to generate the completed suspension
geometry. The main components of the suspension are modelled separately and
linked to each other through the joints. Figure 3-9 shows the front and rear
suspension model in ADAMS/View. To clarity and simplify, the graphics of the
Westfield i-Racer body are modelled as a dummy part that has the same mass and
inertia properties as the body.
a) Front suspension model b) Rear suspension model
Figure 3-9 Front and rear suspension model in ADAMS/View
37
The Westfield i-Racer employs a double wishbone suspension on the front driving
axle and a deformed double wishbone suspension on the rear driving axle. For each
suspension, upper and lower A-shape arms and wheel knuckle are created as parts.
Bushings are used to connect the two A-shape control arms to the sub-frame, and
the control arms are linked to wheel knuckle by using spherical joints. The reason for
using the bushings there is to introduce some steering compliance, so that the
bushings as modelled have extremely high stiffness in the longitudinal and vertical
directions, and are comparatively soft in the lateral direction. Again, all the
components have their own mass and inertia properties.
The spring and damper configurations are created initially using the internal ADAMS
commands so that a spring damper unit can be generated automatically. The springs
are modelled by using a linear single component force that acting between the upper
and lower mounting points of the strut. The force is used to define by a constant
stiffness value with spring deflection. Westfield provided nonlinear damper data that
is modelled using a single component force acting between the same mounting
points as the spring. The nonlinear damping force is defined by a 2D curve with
deformation velocity along the x axis and force on the y axis as shown in Figure 3-10.
Force (N)
Velocity (mm/s)
Figure 3-10 The 2D curve for the nonlinear damper
38
The purpose of the software to model the suspension structure is now well
established and will be discussed further in the next chapter. The results from this
type of analysis are mainly geometric and allow the outputs to be plotted graphically,
such as the camber angle or roll centre position against the vertical wheel movement.
The double wishbone suspension model can be simplified to represent connecting
the A-shape control arms to the sub-frame. Modelling the suspension in this manner
is necessary to calculate the degrees of freedom (DOF) for both the front and rear
suspension systems.
Part 6 6 =36
Spherical 4 -3 = -12
Revolute 2 -5 = -10
Translation 1 -5 = -5
Hook 2 -4 = -8
DOF = 1
Hence, the calculation as shown above is an example showing that the motions of all
six parts at the front suspension have been constrained by using different types of
joints, as shown in Figure 3-11. Note that the type of the joint has to be chosen
correctly in its location to attach the two parts together; otherwise the additional or
incorrect numbers of degrees of freedom may cause the suspension system failure,
or movement in the wrong direction. Also, the revolute joints that connected the
wishbones to the frame will be replaced by bushings when using the full vehicle
model for handling simulations.
39
Revolute Universal
Joints
Joints
Spherical
Joints
Figure 3-11 The parts connected by the correct joints
3.3.3 Steering System
The configurations of the steering system based on linkages and steering gearboxes
are available for both cars and trucks. The steering system in the following sections
is modelled by using a conventional Rack-Pinion system. Modelling a steering
system on the actual vehicle can be represented as shown in Figure 3-12. Basically,
a cylinder is used to model the steering column connects to the vehicle body by
using the revolute joint, which the axis of the joint should align along the line of the
column. The motion or torque inputs of the steering control are applied at this
revolute joint to manoeuvre the vehicle model. Using a translational joint connecting
the steering rack to the vehicle body is represented, and the rack is linked to the tie
rods by using two universal joints. The steering ratio can be implemented into this
system by using a coupler statement that can convert the rotation of the steering
wheel inputs to the translation movement of the steering rack.
40
Figure 3-12 Modelling steering system in ADAMS
When starting the vehicle dynamics analysis, the steering ratio is always an
important parameter for the model design; the relationship between the turning the
steering hand wheel and the steer change at the front wheels has to be known. This
coupler is based on a ratio where one degree of the steering column input generates
the 0.18mm translational motion of the steering rack. This relationship is utilized to
connect the two tie rods at the front suspension, and then control the steer angle at
the wheels. Using a variable statement can measure out the steer angle of the front
wheels. A steering ratio can be determined by a common fashion, degree-to-degree,
which is the 22 degrees of hand wheel inputs relating to 1 degree of front wheel
steering, as shown in Figure 3-13. In both the linkage model and the real vehicle, the
steering ratio would vary when the wheels move in the motion of the full bump and
rebound, or the vehicle rolls.
41
Figure 3-13 The steering ratio for Westfield i-Racer model
3.3.4 Aerodynamic Effects
Some classic assessments of aerodynamics in existing text books (Milliken and

Milliken, 1995; Gillespie, 1992) that corresponds to the vehicle dynamics is
introduced. As is known, when the air flows through the vehicle body, forces and
moments can be produced because of the friction and pressure distribution between
the air and the body surface. Thus, the forces in the longitudinal, lateral and vertical
directions along with the moments in roll, pitch and yaw will arise in the same
reference frame as that of the vehicle body.
In this simulation, the project vehicle moves only in the plane so that it is going to
require at least the formulation of a longitudinal force , a lateral force . The
direction of the project vehicle will be changed during the manoeuvre, hence it is
going to be necessary to model the forces as components in the body-centred axis
system located at the mass centre. The magnitude of the aerodynamic forces can be
calculated out by using the equation 3-2 as shown below:
42
(3-2)
Where = the aerodynamic drag coefficient
= the density of air
= the frontal area of the vehicle
= the velocity of the vehicle in the direction of travel
Using this equation, there are only two forces that were applied within the model to
exert the effects on both the longitudinally and laterally direction of the vehicle. An
expected drag coefficient of 0.4 is used for the EV, and the rest of the data, such as
the frontal area of the vehicle, was included from the Westfield Sportcars. Figure 3-
14 shows that the aerodynamic drag forces act on the vehicle body when turning a
corner.
Aero drag
forces
Figure 3-14 Aerodynamic drag forces in the lateral and longitudinal direction
43
3.3.5 Driveline Modelling
According to the position of the electric motors, the project vehicle is a typical rear-
drive car, and the pair of electric motors is mounted in the centre of the rear axle,
that is, a so-called In-line motors electric vehicle. This type of architecture has been
used in a few designs for racing purposes, such as the Delta E-4 coupe and the Le
Mans Electric. Figure 3-15 b) shows the motors can be directly connected to the
drive-shaft without any transmission system, such as gearbox and differential; thus
the driveline has been shortened and the higher driving efficiency of the powertrain
can be improved.
a) The position of the motors b) The electric motors on the rear axle
Figure 3-15 The motors position of the project vehicle
However, this new driveline will be modelled in this section, as shown in Figure 3-16.
Based on the architecture of the project vehicle with the in-line motors, modelling the
driveline consists of creating two new parts: a cylinder for the twin electric motors,
and two columns for the drive-shafts. And those parts are constrained by using a few
type joints, for example, the cylinder (twin motors) is attached on the vehicle body
using a fixed joint, and linked to the columns (drive-shafts) using two constant
velocity joints. The constant velocity joints are typically used to model the drive-
shafts connections. There are no extra degrees of freedom available on this driveline.
44
Driving Torque
Motors
Drive-shafts
Figure 3-16 Modelling the driveline in ADAMS/View
There are some simulations required to retain the vehicle at a constant speed as in
the following studies, because if the vehicle does not have any form of the driving
torque, it will drive through the manoeuvres at a coasting condition using the
momentum available from the initial velocity. Furthermore, even ignoring the rolling
resistance and the aerodynamic drag force, the momentum of the vehicle will still
lose because of the resistant element of the cornering forces, which is generated at
the tyre during the manoeuvres. However, the means of modelling the driveline is to
implement the driving torques to the driven wheels, then produce the traction at the
contact patch; thus, the points for applying the driving torques on the driveline have
been chosen at the two cylindrical joints. It is more like a real vehicle driven with
electric motors, in which the torque generally outputs from the centric rotor of the
electric motor.
The rotation of the two rear wheels is coupled to the rotation of the rotors of the
electric motors, and the driving torque outputs are equally distributed to each of the
drive-shafts as 50% of the torque. It can be considered as an open differential
mounted on the rear axle, which means the new driveline allows the rear wheel to
rotate at different speeds during cornering, but only by applying the same torque on
each of the driving wheels. Table 3-1 shows all bodies, joints and total DOFs have
45
been calculated by using a function in ADAMS, confirming the model has 15 DOFS
and there are no redundant constraint equations.
Table 3-1 A list of all bodies, joints and DOFs
Before finishing this section to describe how to model the driveline, there is a method
that needs to be represented. Namely, the driving torque acted on the drive-shafts
can be modelled by using an existing TORQUE command in ADAMS/View. Thus,
the torque can be controlled by using a function or subroutine in the TORQUE
command that can be used to apply a control strategy for the vehicle dynamics or
stability control to the driveline. For example, if the model runs at a drifting condition,
the torques from the driveline are deactivated, and the driveline is also able to
produce equal torques on both the driving wheels to represent an open differential.
It is a limitation of an open differential that the amount of total torque applied to both
drive wheels depends on the side with the least traction. For example, when one
wheel acts on a slippery surface, the torque applied to the other wheel would be the
same as that on the slippery side. Thus, the total torque could be reduced and may
be not enough for vehicle propulsion. In additional, the average of the rotational
speed of the two driven wheels equals the input rotational speed of the drive shaft,
and an increase in the speed of one wheel is balanced by a decrease in the speed of
46
the other. These characteristics can be represented by the function or subroutine
associated with the torque to improve the control strategy of the driveline.
Furthermore, even a more complex control strategy, such as torque vectoring control,
can exploit this approach to implement all the control concepts or requirements into
the driveline, and that will be discussed in Chapter 6.
3.4 Tyre modelling
The subject of extensive research, tyre modelling mainly focuses on how to model
the forces and moments generated at the contact patch where the tyres touch the
road surface. Pacejka and Sharp (1991) provided and developed the most common
tyre model, which have been broadly used, and the tyre model can compromise the
modelling of the tyres between the accuracy and complexity.
The research mainly focuses on a vehicle handling study, thus the project model is
generally manoeuvring on a flat road surface. In general, the tyre model is used to
create and model the forces and moments generating at the contact patch between
the tyre and the road surface. These forces and moments will transfer through the
suspension into the vehicle body. There are the three orthogonal forces and the
three orthogonal moments that can be calculated by the tyre model at each wheel
centre, controlling the vehicle body motion. Again, for the handling of the analysis,
the calculation of the tyre model is mainly based on the longitudinal direction (driving
and braking forces) and lateral forces, also including the formulation of rolling
resistance and aligning moments. A flat 2D road file is used as a part of the model;
furthermore it has the function to change the friction coefficient of the road surface to
represent the various surface textures, namely dry, wet and ice conditions.
3.4.1 Modelling virtual tyre rig
The aim of modelling the tyre test rig is to provide a tool to integrate a given tyre
model and its associated data set before using this with a vehicle simulation.
47
Figure 3-17 The tyre test rig model
The modelling of the tyre test machine is illustrated in Figure 3-17 and includes a
tyre model that rolls forward on flat surface, representing the actual tyre interaction
with a moving belt in a real experiment. In this model, the tyre model is attached to a
test rig model using a revolute joint, the axis of which is aligned with the spin axis of
the wheel. A rotational motion can be applied to a cylindrical joint (aligned vertically)
for which the rotation represents the side-slip angle on the tyre during the simulation.
A vertical force can also be applied to the carrier part to represent the tyre load. For
the parts and the joints used in this model, the overall system has only two degrees
of freedom. These represent the spin motion of the tyre and the vertical movement of
the wheel centre.
3.4.2 Tyre model for project vehicle
The tyre model used in extensive computing-based simulation is highly important

when analysing dynamic behaviour. Because the Westfield Sportcars company
cannot provide the particular parameters for modelling the given tyre as it is used on
the project vehicle, an alternative approach involves an initial trial with the Fiala tyre
model (Fiala, 1954). As is known, there are only ten input parameters that are
required for creating the Fiala tyre model; this advantage can be directly correlated
48
to the characteristics of the tyre. Nevertheless, the major limitations of the tyre model
include the following: a) the model does not have the ability to represent the
combined situation, such as cornering and braking, or cornering and driving at the
same time. b) Lateral force and aligning moment cannot be calculated as having a
camber angle. c) When the tyre with the vertical load moves on the straight line, the
variety of the cornering stiffness is not modelled.
Another suitable alternative approach is established to fulfil the requirements of the

simulation. It is achieved by using the Magic Formula tyre model (Bakker et al. 1986,
1989), of which the Magic Formula tyre model is now most well-established, based
on the work by Pacejka (Pacejka and Bakker, 1993). This model is a point follower
model, which uses an empirical mathematical form that can be shaped with a
relatively small number of parameters to fit more or less any measured tyre dataset.
The point follower term refers to the fact that the tyre is idealized as a single point
of contact, which has aggregate forces and moments for the entire contact patch.
This is the way that data is recorded on a typical tyre test machine and is completely
suitable for smooth road handling calculations at low tyre slip angles. Because of the
absence in the particular coefficients of the tyre model for the project vehicle, a set of
standard data was used, as shown in Table 3-3. The coefficients need to be adjusted
in a reasonable region within which the results from testing the tyre model on the rig
can fulfil the expectations of this particular tyre model under the dynamic
manoeuvring. Then, the tyre model has been tested on the virtual tyre test rig in
ADAMS, and examination of the tyre results will be shown in the following section.
49
Table 3-1 A set data for tyre model
These coefficients of the tyre model will be used for all simulations in later Chapters.
For these simulations, the tyre model was steered in a region of the slip angle and
operated under four different vertical loads (1962N, 3924N, 5886N and 7848N). The
examination of the tyre results is shown in Figure 3-18 and Figure 3-19; the results
represent some curves for the lateral force and the self-aligning moment during the
changes of the tyre slip angle. The results show that the tyre performed in an
intended behaviour; nevertheless, the coefficients created in the tyre model were
only modified to an approximate level. Further validation of the coefficients is needed
to ensure that the data from the test rig model are as near the same as possible to
the actual tyre characteristics, which has to be taken into account when examining
all of the results generated from the simulations.
50
Figure 3-18 Lateral Force Vs Slip Angle tested on the Virtual Tyre Test Rig
Figure 3-19 Self Aligning Moment vs Slip Angle
3.5 Conclusion
The project vehicle has been represented in detail, with the entire architecture of the
given electric vehicle with RWD being included, and the components and the
configuration of the new driveline also being introduced. The capacity of the twin
electric motors could fulfil the requirements of the racing vehicle. The lightweight
chassis and the wishbone suspension for the project vehicle were described.
51
Several sub-systems of the vehicle, for example the vehicle body, the suspension
system and the driveline, have been modelled in ADAMS, and the audit was carried
out to ensure the reliability of the model. Also, the aerodynamic effects were
considered in this section. The driveline was able to produce the variable driving
torque at the wheels, so that it could represent the different driving modes, such as
the open differential and the torque vectoring control. The Magic Formula tyre
model was added on the vehicle model, and also used with the virtual test rig model
to examine characteristics of the tyre model in the intended behaviour due to
modifying the coefficients of the tyre model.
52
4 Measurement and analysis of the virtual model
4.1 Introduction
When building a vehicle model in a computer-based environment, the accuracy of

the simulation results relies on the accuracy of the model and the vehicle parameters
used to build the model. Hence, there are several methods available to validate the
vehicle model, such as kinematic studies and dynamic manoeuvres.
In order to verify the performance of the suspension system as modelled in Chapter

2, a range of characteristics will be determined through simulation of a quarter
vehicle model. In this chapter it will be shown that the full vehicle model will be
analysed in a number of ways that will provide information to support the following
investigations. Also, the steady-state cornering manoeuvre will be used to define the
basic driving characteristics of the vehicle model.
4.2 Kinematic analysis
The full vehicle model has been created in the previous chapter, where the
arrangement of the suspension system consists of the typical linkages for the
double-wishbone suspension to connect the wheels to the vehicle body. The
interaction of the linkages restricts the wheel plane to undergoing combined
translation and rotation. These motions are normally determined with respect to the
vehicle body, so the following descriptions will explain the kinematic analysis of the
wheel. For example, toe change for suspension linkages directly steers the front
wheels and generates lateral force and yaw moment. Camber angle is a
compensation angle to maintain the tyre perpendicular to the road surface when the
vehicle body rolls. Half-track change influences the lateral velocity of the tyre contact
patch via the roll rate. Hence, the slip angle of the tyre is affected, since the angle is
defined as the arctangent of the lateral and longitudinal velocities; an increase in the
lateral velocity directly increases the slip angle.
53
4.2.1 Quarter vehicle modelling
In this section, the basic function of the suspension system is represented from a
functional perspective. The front suspension of the vehicle model has been built as
one of the best-known suspension types, the so called Double Wishbone system,
and also the double wishbone system has been created for the rear suspension.
Using the software to analyse suspension geometry is now well established and will
be discussed further in the next section of this chapter. The output from this type of
analysis is mainly geometric and allows results such as the camber angle to be
plotted graphically against the vertical wheel movement.
Modelling the quarter vehicle is discussed by using the existing double wishbone
suspension system at the front axle. Basically, the suspension model uses the
revolute joints to connect the upper and lower A-shape arms to the chassis as shown
in Figure 4-1. Imparting a vertical motion to the suspension is achieved through a
jack part that connects to the ground using a translational joint. Therefore, at the
translational joint the applied motion moves the jack through a range of vertical
movements that assumes the suspension moving between the bump and rebound
positions. An in-plane joint is applied to link the jack to the wheel knuckle where the
wheel centre is located. In addition, this joint only constrains the wheel centre at the
top of the jack to move in the plane, but the wheel can still rotate or move in the
lateral and longitudinal directions.
Figure 4-1 Modelling the front of the quarter project vehicle
54
Note that the resulting system model has a total of zero degrees of freedom. The
movement in the system is due to the introduction of a user prescribed time
dependant motion that accounts for one degree of freedom constrained from the
system. The resulting analysis is kinematic. The introduction of one or more
additional degrees of freedom would result in a dynamic analysis.
4.2.2 General approach for kinematic analysis
For a suspension system, the main function of a multi-body systems model is to

simulate the geometric position and orientation when the suspension has a vertical
movement between the full bump and full rebound positions (Blundell, 2004). When
simulating the vertical motion, the output is only suitable for a kinematic or quasi-
static analysis. As mentioned, the motion applied to the translational joint moves the
wheel between the 60mm rebound and 60mm bump positions. This results in
particular times for the movement, at 0.25 second for bump and at 0.75 second for
rebound, which enables the present of the continuous and smooth animation of the
movement cycles. The duration of the simulation time is 1 second with 100 output
steps, which ensures the results are calculated over the full bump and rebound
range. In ADAMS, a motion statement was used to move the translational joint
during all the simulations.
The motion statement is shown below, where the total movement between bump and
rebound is 120mm:
55
a) Front suspension b) Rear suspension
Figure 4-2 Kinematic simulation for suspension system
Something needs to be clarified at this stage in which the variable converts the time
in seconds to degrees in the function that represent one cycle over 1 second during
the simulation time. If implementing an asymmetric movement which involves
different distances in the bump and rebound, the motion input needs a more complex
function to operate in ADAMS.
4.2.3 Suspension measurements
Since ADAMS/View is general purpose multi-body dynamics simulation software, it

does not have any specialized tools to measure the various vehicle parameters that
need to be monitored and recorded to compare the performance of the vehicle.
Hence the different variables to measure and monitor various parameters such as
steering angle, camber angle and roll angle are created. The descriptions provided
here will be limited to the most commonly calculated outputs. The outputs shown in
Appendix 5 are for those plots that are normally used to compare simulation results
with physical test data for validation of the virtual model in ADAMS.
Bump Movement: Measuring the wheel movement upwards in the positive

vertical direction relative to the vehicle body.
Wheel Recession: Measuring the wheel movement rearwards in the positive

longitudinal direction relative to the vehicle body.
56
Half-Track Change: Measuring the wheel movement outwards in the positive lateral
direction relative to the vehicle body. All of these are shown in
Figure 4-3
This item has been removed due to third party copyright. The unabridged version of the
thesis can be viewed at the Lanchester Library, Coventry University.
Figure 4-3. A model diagram for the WR, HTC and BM (Blundell. 2004)
Camber angle: This measurement uses two markers that are positioned at the
wheel knuckle. The first marker (angle_wc) is situated at the
wheel centre on the knuckle, and the second marker (angle_sa) is
positioned on the axis of rotation of the wheel. The measure is
defined in ADAMS by the equation as shown in Figure 4-4:
Wheel Steer Angle: The road wheel steer angle is calculated using the same
markers used to measure the camber angle. The steer angle
measure is defined by the equation as shown in Figure 4-4:
57
Figure 4-4. A diagram shows camber angle and steer angle
Castor Angle: The measurement of the castor angle is in degrees in the side
elevation between the vertical and the steering axis. The points for
measurement are located at the upper and lower ball joints of the
suspension system connect. Thus, the angle is defined by the
equation as shown in Figure 4-5:
Suspension Trail: The suspension trail is the length in the longitudinal direction
between intersection of the steering axis and the ground, and the
wheel base which can be described as below:
58
Figure 4-5. Calculation of castor angle and suspension trail
Steer Axis Inclination: The steer axis inclination is that the angle is measured at the
front elevation between the vertical and the steering axis
(from the upper ball joint to the lower ball joint). It can be
explained by the equation below (see Figure 4-6):
Ground Level Offset: The ground level offset is the length in the lateral direction
between intersection of the steering axis and the ground,
and the wheel base (see Figure 4-6):
Figure 4-6 Calculation of steering axis inclination and ground level offset
59
4.3 Dynamic analysis
In the automotive industry, there is considerable focus on ride and handling where
multi-body systems analysis is deployed to support design and analysis work. There
are multiple sources of ride vibration which generally fall into two areas, namely road
roughness and on-board sources. For vehicle dynamic analysis, the best starting
point is to know the basic properties of a vehicle and its suspension system i.e. the
motions of the body and axles. The body as a portion of the vehicle moves as an
integral unit on the suspension. The suspension and wheels, as the un-sprung
masses, move as a rigid body and impose excitation forces on the sprung mass.
Thus, one must look into structural modes of vibration and resonances of the sub-
system on the vehicle.
Before starting the computer-based simulation for dynamic analysis, manual

calculations are necessary to introduce how to find the natural frequencies of the
vehicle body and un-sprung between the suspension spring and the tyre spring. A
quarter vehicle model with two degrees of freedom and the data to support the
calculations is shown in Figure 4-7.
This item has been removed due to third party copyright. The
unabridged version of the thesis can be viewed at the
Figure 4-7 Two degrees of freedom quarter vehicle model (Blundell, 2004)
The un-damped natural frequencies for the body and for the un-sprung mass
can be estimated using the following equations. Note that for the body an equivalent
stiffness is determined to represent the combined contribution of the road and tyre
springs:
60
4.1
4.2
4.3
The modal solution uses numerical perturbation methods to estimate mass and
stiffness matrices about an operating point before solving the eigenvalues in the
normal fashion. For vehicle design, modal analysis is utilized to measure and
analysis the dynamic response of vehicle structures when excited by multiple inputs.
4.3.1 Total degrees of freedom
Having introduced the modelling of the full vehicle model described in Chapter 3, the
total degrees of freedom for this model need to be determined in the following
section. For example, the vehicle body is a free floating rigid body in three-
dimensional space so that it has six degrees of freedom. The vehicle body for the
dynamic simulations does not have any connections directly to the ground, and is
only connected to the suspension. To determine and understand the total degrees of
freedom in the system, the Gruebler equation can be used as shown below:
4.4
For the vehicle model, it is necessary to verify the model size in terms of the degrees
of freedom, and the calculation of the number of the DOF is based on the Greubler
equation. Table 4-1 shows there are 6 degrees of freedom for the vehicle body, each
of the four suspensions has only one translational degree of freedom in the vertical
direction and one rotation degree of freedom exists for the wheel parts. This makes a
total of 15 degrees of freedom for the entire model. Moreover, due to the model
using the bushes on both the upper and lower wishbones, more DOFs are
61
introduced for each suspension system. Thus, the total DOFs of the full model is
increased to 55.
Table 4-1 Total degrees of freedom for the model
a) Without bushes b) with bushes
44
55
4.3.2 Dynamic model
In this case study, a dynamic model was established as shown in Figure 4-8, and it
is considered good practice to examine the dynamic model for its eigenvalues to find
the Natural Frequencies of the sprung and the unsprung mass. The model is
assembled into its Ride Rig configuration and the motion of the input rigs is set to
zero. The approach used for the calculations in ADAMS is the so-called small
perturbation method in which a small disturbance is introduced at each degree of
freedom in order to construct stiffness and mass matrices, which are then solved in
the traditional manner.
62
Figure 4-3 The dynamic model for modal solution
The dynamic model is essentially a basic version of the full vehicle model which only
includes the suspension system and the vehicle body, thus reducing the total DOF to
10. The four tyres are replaced by a single spring unit that has the same stiffness as
the tyre, and in-plane joints are added between the spring unit and the ground part to
ensure that the unit is perpendicular to the ground, and also includes a planer
constraint ( direction) on the ground. The simulation runs in the dynamic mode for
two seconds and examines the vehicle model to find the natural frequencies for each
degree of freedom.
The results are shown in Table 4-2. The information includes: the sequential number
of the mode that is predicted by the Eigen solution, natural frequencies
corresponding to the modes, damping ratios for the modes and list the real and
imaginary parts of the eigenvalue. The table of results from ADAMS/View produces
ten modes of vibration for this full linkage vehicle model, and also includes the total
number of the parts and the joints. Finally, the model is verified to show there are ten
degrees of freedom for the model and no redundant constraint equations.
63
Table 4-2 Display of the eigenvalues in tabular form
The calculation for eigenvalues and eigenvectors can be done by using a function of
ADAMS which can give a better understanding of the models natural frequencies
and mode shapes. After the calculation is complete, ADAMS can display in tabular
form the Eigenvalues, in which the particular stability behaviour of the model
depends on the existing real and imaginary parts of the eigenvalues, and the signs of
the real parts and the different values of both parts. However, the model is a stable
system in which all the real values are negative and the system has an imaginary
part in which one pair of the values corresponds to each mode. In addition, each of
the natural frequencies corresponds to the mode shapes that are the deformations of
64
the sprung mass and the un-sprung masses. Some explanations are shown in Table
4-3.
Table 4-3 Explanation of the natural frequencies at each mode
Mode Frequency Comment

1 0.0398 Hz Body on rig stabilisation springs - Yaw
2 0.086 Hz - Fore-aft
3 0.178Hz - lateral
4 1.408Hz Primary Ride 1 Heave
5 2.478 Hz Primary Ride 2 Pitch
6 3.527 Hz Primary Ride 3 Roll
7 17.747Hz Rear left wheel hop
8 17.875 Hz Rear right wheel hop
9 18.94 Hz Front left wheel hop
10 19.144Hz Front right wheel hop
Moreover, in ADAMS the first six modes can be represented by animations of

vibration for the vehicle body supported by the suspension systems as shown Figure
4-9. The animations give a better understanding of the response for the vehicle
body at primary ride, for example, the natural frequency 0.0398 Hz at the first mode
is calculated when the vehicle body is yawing. Reading the rest of the animations as
shown below, the body movement and the natural frequency at each mode can be
easily understood. However, the highest natural frequency of the sprung mass,
around 3.5 Hz is found in the roll mode because the vehicle body is modelled as a
single rigid body. This simplicity may give a large number for the torsional stiffness of
the vehicle body structure in the roll mode. As is well known, the stiffness of the body
structure is required for vehicle handling and drive performance; likewise, its effects
on safety requirements are a major issue in the vehicle industry.
65
a) Mode_1 Yaw b) Mode 2 Fore-aft
c) Mode 3 Lateral d) Mode 4 Heave
e) Mode 5 Pitch f) Mode 6 Roll
Figure 4-4 Animations of the modes at primary ride behaviour
66
In this model, the suspensions and the rigs are the un-sprung mass of the vehicle
that is the second largest of the masses. In general, all un-sprung masses have the
vertical hop mode that is excited by the road uneven inputs adding to the vibrations
present on the vehicle. Thus, the four-wheel hop modes are sensibly positioned with
respect to each other, and are represented in Figure 4-10. It can be seen that the
natural frequencies are much higher than the sprung mass resonance, so that the
sprung mass remains stationary during the wheel hop.
a) Wheel hop - Rear left b) Wheel hop Rear right
c) Wheel hop Front left b) wheel hop Front right
Figure 4-10 Animations of the modes at wheel hop behaviour
67
For a normal driving vehicle, the calculated frequency will be approximately 10 Hz.
Friction in the suspension will increase the effective spring rate for small ride motions
which in turn will increase the frequency to 12-15Hz. However, the wheel hop
frequencies are high compared to those of the normal passenger car due to the
track-racing nature of the vehicle, which means the vertical tyre stiffness and spring
damper face are comparatively high, raising the wheel hop mode frequency. The
purpose of kinematic and ride calculations are to check the virtual model has been
built correctly using a Multi-body system for following analysis, i.e. the correct
numbers of parts and the types of joints are being used in the model. Overall, the
model behaves well dynamically and can be trusted for its intended 0-20Hz dynamic
use in a numerical sense. However, in terms of representing real vehicle behaviour
there is one important caveat the roll stiffness distribution as modelled may be very
different to the real roll stiffness distribution due to vehicle frame flexibility, about
which no information was available.
4.4 Steady-State Handling analysis
The classical treatment of the behaviour of the vehicle model, that is called steady
state cornering, is described out in this section. Steady state means the vehicle
states are unchanging with time and speed as the vehicle travels on a constant
radius circle. This treatment will be described in more detail; running the model
around a circle with constant radius at a range of constant speeds, the increments in
the speed should correspond to the increments in the lateral acceleration. Moreover,
this approach is very practicable to perform and so it will be used for the basis of
vehicle dynamic analysis in the following discussion.
4.4.1 Driver behaviour modelling a path following controller
Before starting the treatment, the first consideration of modelling the steering inputs
is to represent the driver as part of the full vehicle system model. In this research,
68
the inputs to the steering wheel are designed with both open-loop and closed-loop
control. The steering input with the open-loop control requires a time-dependent
rotation that will be applied to the steering system. If applying a closed-loop control
to the steering system, the aim is to adjust the input to produce the desired trajectory.
However, the aim of the control method is to minimize the error that is the difference
between the actual output and the desired output. In the literature, there are a variety
of controller models suitable for modelling driver behaviour in existence. Some, such
as ADAMS/Driver developed as part of the MSC.ADAMS modelling package, are
very complete others, such as the two-loop feedback control model are simpler
(Blundell and Harty, 2004). Some researchers prefer to use a preview distance for
controlling the trajectory of the vehicle, with an error of lateral deviation from the
intended path. However, there is usually a difficulty associated with this since the
lateral direction must be defined with respect to the vehicle. For normal driving this
type of model can produce acceptable results but for manoeuvres such as the ISO
3888 Lane Change the behaviour becomes unacceptably oscillatory particularly after
the manoeuvre. An alternative method, used by the authors with some success for a
variety of extreme manoeuvres, is to focus on the behaviour of the front axle. This
model fits with drivers experience of driving at or near the handling limit, particularly
on surfaces such as snow where large body slip angles highlight the mechanisms
used in the drivers mind.
Ground Plane Velocity
The path following controller model focuses on the behaviour of the front axle in
which the principle of the control method is to determine the yaw rate error between
the No-slip yaw rate of the front axle and demanded yaw rate. The demanded yaw
rate is the driver controlling the vehicle to follow the path as expected at a range of
speed. The formulation used is described below. All subscripts x and y are in the
vehicle reference frame.
In Figure 4-11, the ground plan velocity Vg is the given from the components and
using
69

4.5
Figure 4-115 Explanation of Ground Plane Velocity
For the reference frame used here and as the vehicle moves forward, the signs of
and are negative. A variable command is used to calculate within the model to
find out what the significance of the ground plane velocity is. The units of model work
in and will be set up in . As the simplest start, the model will run for the
steady state cornering, the driving torques will not act on this model and it only has
an initial velocity on all parts, assuming the vehicle is in a coasting condition.
Demanded Yaw Rate
The demanded yaw rate d is found from the forward velocity and path curvature
using:
4.6
4.7
70
Figure 4-12 Explanation of Demanded Yaw Rate
For example, if the model runs at a 33m constant radius circle as shown in Figure 4-
12, the path curvature is = 0.0303m-1. So the demanded yaw rate can be easily
calculated depending on the ground plane velocity of the model.
Path curvature and path length
The path curvature for a circle should be a constant value, but it needs some
transition curves with varied radiuses to link path curvature from the straight line to
the constant radius circle. This approach can avoid abrupt steering inputs while the
vehicle model follows the desired path. The desired path curvature can be converted
to a SPLINE command as used in ADAMS, shown in Figure 4-13. The X values are
path lengths depending on vehicle speed, and Y values are path curvatures. Note
that the value of the path curvature can be positive or negative, so it needs some
definitions before the simulation. In this cornering scenario, when the path curvature
is set up for a positive value the vehicle model turns to the left and the negative
value makes the model turn to the right.
71
Figure 4-6 Path curvature for steady state cornering
However, the path curvature is determined and in order to simulate the model for a
range of speeds, the path length corresponding to the different speeds is required.
The path length is found by integrating the ground plane velocity of the vehicle to
give a distance-travelled measurement. Using this measurement, the path curvature
can be surveyed in the model.
Body Slip Angle
The explanation for the body slip angle, which appeared in Figure 4-22, can be found
from the velocities and using:
( )
4.8
A variable statement is built to convert the formula into ADAMS, thus the body slip
angle of the vehicle body can be calculated when the model is cornering. The
numerator requires the lateral velocity of the body and the units set up in m/s. In
72
order to avoid having the denominator equal to zero, a correction factor has been
added into the variable statement that can avert any the numerical failure. The effect
of the factor is negligible.
Centripetal Acceleration
First, the variables of lateral acceleration and longitudinal acceleration are

created and calculated in the vehicle body axis system, the units set up in m/s 2.
Then, the centripetal acceleration is found from the components of acceleration
and using
4.9
The Front Axle No-Slip Yaw Rate
The front axle no-slip yaw rate fNs is found from the components of the centripetal
acceleration , the yaw acceleration , the distance , from the mass centre to the
front axle and the ground plane velocity using
4.10
Yaw rate error
The yaw rate error is then found from the demanded yaw rate and the front
axle no-slip yaw rate using
4.11
73
4.4.2 Driver behaviour modelling a survey controller
Before building the survey controller, some basic knowledge of the vehicle driving
characteristics during cornering needs to be mentioned. At the lowest speed, the
performance of the actual vehicle corresponds very closely with the geometric yaw
rate of the vehicle body. As vehicle speed rises, the lateral acceleration of the
vehicle is increased so that the tyres must develop lateral forces, and slip angles are
presented at each wheel. Due to the frictional limitations of the tyres, the vehicle
cannot reach the geometric yaw rate for a large steering input. Thus, the
characteristics of the vehicle are designed to have further modifications in which the
yaw rate gain should be reduced even when the tyres are not saturated. Therefore,
the over- or under-steer can be easily defined by the ratio of the geometric yaw rate
and the actual yaw rate, i.e. if the ratio is greater than one, the vehicle is under-
steering and when the vehicle is over-steering, the ratio is less than one,.
As is known, the tyre cornering force curve is treated as having linear, transitional
and frictional regions, and the project racing model may operate in all of these
regions. At the starting point, the response of the vehicle related to the driver control
will be operated in the linear region, corresponding to the lateral acceleration of the
vehicle body, which is about 0.3g (Milliken, 1995). Thus, a similar control method as
the path-following controller is used in the survey controller model in which only the
target behaviour differs slightly. The survey controller model still focuses on the
behaviour of the front axle, but the aim is to control the lateral acceleration of the
model at its linear region. Therefore, the target is set up for the body lateral
acceleration at 0.3g so that it can be converted into ADAMS by using the SPLINE
command. The SPLINE command represents the Y axis as target lateral
acceleration depending on the X values for simulation time, and the units are
converted in m. Finally, the lateral acceleration can be found by using:
4.12
74
4.4.3 Steering torque input
The closed-loop driver model with the Path Following and Survey controllers will
be applied to the vehicle model by using a typical PID controller. As one of the
control technologies represented in the literature, the PID controller has the
advantage by which it produces continuous output and there are no steps that are
quite like the behaviour of a real driver. For example, when using the driver model
with the path-following controller, the yaw rate error is minimized in three ways: a
control effort is applied in proportion to the error, and then the error can be integrated
and also differentiated. In the PID formulation, the integral and derivative terms can
be represented by using DIF statements and variables in ADAMS. The controller
theory section for Yaw Rate Error (YRE) is shown below
4.13
As stated, the simple steering system as modelled in the previous Chapter is using
the steering rack connecting the steering column and the tie rods of the front
suspension. The rotational motion of the steering column is related to the
translational motion of the steering rack through the coupler joint. This coupler is
based on a ratio where a degree of the steering column input generates the 0.18mm
translational motion of the steering rack. Therefore, the implementation of the control,
as shown in Figure 4-28 is to apply the torque to the steering column and the driving
of the vehicle model follows the desired path.
75
Figure 4-7 Steering torque acting on steering wheel
When the driver model implements the control signal on the steering wheel, the PID
controller can modify the steering torque to minimize the yaw rate error. This is more
like the behaviour of a real driver because when the yaw rate error increases the
driver will apply more torque on the steering wheel to reduce the error and keep the
vehicle on track. The SFORCE statement in ADAMS is used for implementation of
the steering torque input as shown in Figure 4-29.
Figure 4-8 Steering input for the path following model
76
4.5 Simulation results
4.5.1 Steady state cornering behaviour
The cornering behaviour of a vehicle is an important performance attribute that

normally corresponds to the handling. There are three phases when a vehicle makes
a turn into a corner, turn entry, steady-state cornering and turn exit. During the
second phase, the radius of the path and velocity of the vehicle are constant. The
Ackermann angle which is the average steer angle of the wheels, =L/R, can be
found.
In the general steady state cornering manoeuvre, it is necessary to consider the

relationships in Figure 4-30. The under-steer region requires more steer angle than
the Ackermann angle to drive the vehicle in the desired circle. Similarly, the over-
steer region has less steer angle compared to the Ackermann angle.
Figure 4-9 The explanation of Under-steer or Over-steer (Gillespie, 1992)
To determine the characteristics of the project vehicle, such as under-steer or over-

steer, the constant radius turn test procedure, which is based on the British standard
(ISO 4138), can be used. The procedure may be summarized as follows: for the first
step, the model starts running at low speed to find the Ackermann angle
77
(86.3degree). Then, increasing speed in steps produces increments in lateral
acceleration of typically 0.1g, and measures the steering inputs when the model runs
in steady state cornering at each speed. Finally, the measurements can produce a
graph as shown in Figure 4-31 that can indicate the project vehicle is understeering
when turning the corner, and it also can provide the under-steer gradient .
100
Steering wheel angle (deg)
80
60
40
20
0
0 0.2 0.4 0.6 0.8
Lateral acceleration (g)
Figure 4-31 Determination of the project model
4.5.2 Path following behaviour
An initial comparative analysis of the actual and demanded driving behaviours is

exhibited. As Figure 4-32 shows, the classical treatment based on steady state
cornering verifies that the model can exactly follow the desired path during cornering
by implementing the controlled steering torque on the steering column.
Figure 4-32 The model runs at the steady-state cornering
78
Moreover, the implementation of the closed-loop driver control with the PID controller
provides the adjustable steering torques so that the yaw rate error can be minimized
as shown in Figure 4-33.
Figure 4-10 Demanded Yaw Rate Vs Front Axle No-Slip Yaw Rate
4.5.3 Vehicle linearity
In a process similar to that for the driver model with path following controller, the
survey controller operates the vehicle model running at its linear region as shown in
Figure 4-34. The linear steady-state control response characteristics are given as a
series of ratios: curvature response, yawing velocity response and lateral
acceleration response. As mentioned, the reason for steering the vehicle is to
change the direction of the vehicle by developing the yaw rate. Thus, the notion of
yaw rate gain which is a ratio of yaw rate to steer angle at the front wheels, becomes
useful to describe the yaw velocity response to control inputs.
79
Figure 4-11 The vehicle lateral acceleration versus target lateral acceleration
Again, for vehicle stability and control, the basic characteristics of the vehicle need to
be considered. If the response of the vehicle is less than what might have been
expected, the term understeer is used. When the vehicle yaws more than expected,
the term oversteer is used the response of the vehicle exceeding what might have
been expected. So far the approach for measuring the steering inputs to define the
under- or oversteer has been discussed. Remaining with Newtonian friction to
describe the behaviour of the tyres, one further fundamental point is worth
establishing, which is the relationship between the yaw rate gains with forward
velocity as shown in Figure 4-36. When the model is at low speed, the yaw rate gain
is very close to the ideal yaw rate gain value, which means the response of the
vehicle control is more like that of a natural steer vehicle. When the vehicle speed
increases, the vehicle is unable to achieve the ideal yaw rate for large steering
angles. The characteristics of the vehicle are performed to further reduce the yaw
rate gain even when the tyres are not saturated. Furthermore, based on the vehicle
characteristics, an authority envelope of torque vectoring control is carried out in
Figure 4-36, so that the aim for the torque vectoring control which will be developed
in this research is to control the vehicle behaviour close to its own characteristics
when running at various conditions.
80
Ideal Yaw
Rate Gain
Torque
Vectoring
Authority
Envelope
Actual Yaw Rate

Gain
Figure 4-12 Yaw rate gain versus vehicle speeds
4.6 Conclusion
A summary of the simulation results was collected from these investigations to

evaluate the overall characteristics of the project model during the different analyses.
First, the characteristics of the suspension system have been determined by the
kinematic analysis in which the results can be compared with the laboratory data
from manufacturers to validate and refine the computer-based vehicle model.
Second, the dynamic analysis proved the model is a stable system with correct
degrees of freedom, the natural frequencies at each mode being calculated and the
animations clearly explaining the movement of the sprung and un-sprung mass. Also
the influences of the natural frequencies at each mode were analysed.
Finally, the vehicle model ran for the characteristic driving manoeuvre - steady state
cornering, and implementing the driver models with path following and survey
controller to the full vehicle model was represented. Of most importance is the
characteristic of the project vehicle which verified that the project vehicle is an under-
81
steer vehicle. Furthermore, Figure 4-23 for the yaw rate gain against the vehicle
speed indicated that the aim of the further control strategy for active yaw control is to
overcome all the variations, and then to dominate the vehicle running at its particular
characteristics.
82
5. Comparison of architecture for electric vehicle
5.1 Introduction
For conventional vehicles, a main difference in architecture is the position of the

Internal Combustion (IC) engine which can be located at the front, middle or rear of
the vehicle. The type of mechanical driveline also is another important factor of
vehicle architecture, such as Rear Wheel Drive (RWD) or All Wheel Drive (AWD). A
novel architecture for electric vehicle driveline was a design that includes freedom to
move the motors to a particular space in the vehicle, for example, mounting the
motors within individual wheels. In addition, the electric motors can also be located in
the middle of the chassis at the front, rear or both axles.
In recent year, the architecture of electric vehicle with in-wheel motors on the market
has high driving efficiency, thus any other architecture may be neglected. Westfield
Sportcars, a producer of in-line motors, has recommended a series of wide-ranging
studies into the vehicle stability and driveability performances. Therefore, a
comparison will be discussed in this chapter that includes ride comfort check and
drivability check by using the ADAMS model. The discussion can provide a
comprehensive overview of using in-wheel and in-line motors in this particular racing
model.
5.2 Architecture of in-wheel motors vehicle
The change in the un-sprung mass significantly affects the ride and handling
behaviour thus it is an important parameter. Many feasibility researches have been
done broadly for in-wheel motors, some specific and detailed measures for the sizes
of the effects in dynamics behaviours have been taken. The ride and handling
performance from subjective and objective measures suggests that the modern
development toolbox is easily capable of restoring dynamic performance. However,
the advantages of the in-wheel motors in terms of packaging and vehicle dynamics
control are of substantial interest to the vehicle dynamics community (Anderson and
Harty 2010).
83
In recent years, a popular novel architecture of electric vehicle has developed by
using electric motors mounted in driving wheel hubs that is effective in reducing the
components of the entire traditional drivetrain. The higher driving efficiency of the
powertrain can be improved which the fuel tank to wheel efficiency of the
conventional powertrain is about 20% (Gschel and Burkhard, 2008), that of an
electric vehicle with in-wheel motors amounts approximately 80% (Neudorfer, H and
Binder, A, 2006).
Some new designs of suspension system with in wheel motor were published in last
decades. First, Willberger and Ackerl (2010) designed an elementary model of
intergraded suspension with in wheel motor as shown in Figure 5-1, which consisted
of an electric motor, gear box and brake system. Eduardo and Rojas (2011)
developed a single wheel suspension system to integrate entire demanded
components together. The integration mainly considers in two aspects: first, if an
electric motor will be integrated into an existing suspension system, the effect on the
behaviour of the characteristic suspension parameters has to be minimized during
the adjustment should reach the target behaviours of the vehicle handling. Moreover,
it enables preventing any impact between suspension elements and the in-wheel
motor. The modification bases on an original rear axle of Ford Focus 1999
suspension system, as mentioned, an electric motor and a gearbox are equipped
into the wheel hub as the practical application example.
(a) Package design: 1. Electirc (b) Optimized suspension system for

motor 2. Gearbox 3. Brake passenger vehicles with in-wheel
disc 4. Suspension arm motors (Eduardo and Rojas,
(W illberger and Ackerl, 2010) 2011)
Figure 5-1 Different design of in-wheel motors
84
There are several different types of in-wheel motors in the literature mainly used for
either pure and hybrid electric propulsion units in vehicles. The aim is to satisfy
different demands of workloads depending on the uses of the vehicle. For current
technologies, there are two motor types as asynchronous and synchronous
machines can represent a realistic electrical drivetrain system due to assembly and
cost reasons. The former has its advanced characteristics - robustness, simple
construction and low costs is one of the most widely used electric motor and due to
the compact design that the latter also has steadily developed in last decades.
Willberger and Ackerl (2010) have done some comparisons in which they indicated
that conclusively from a constructive point of view the synchronous motor has better
characteristics concerning an application within an on-demand wheel hub scenario.
Furthermore, choosing the correct type of electric motor as used in the wheel hub
depends on the velocity of vehicle, which the subsequent values can define the
amplitudes of traction force acting on the driving wheels. The approach derived a
prior analysis of different load requirements, that the desired torque has significant
influence on the design of the in wheel motor. In fact, the in wheel motor has been
required with high driving torque in a limited space, that means the in wheel motor
may be redesigned due to individual diameter-length of the driving wheels for a given
design torque. Moreover, the considerations under the geometric restrictions of the
wheel hub should involve the viable gear radios, compact design of electric motor
mounting in the wheel hub with the existing cooling issue and the braking system
needs to modify in order to integrate the electric motor with a gearbox.
Figure 5-2 Protean Electric in-wheel motor
85
A more complex design is developed by a producer Protean Electric, where the
packaging is complicated by fitting a brake that the concept removed the traditional
braking system in the wheel replacing by an inside out disc clearly shown in Figure
5-2 to preserve the track width and structural integrity of the suspension system. Due
to a smaller rubbing surface of the brake disc and reducing bending moments, twin
callipers are used to allow coupled friction forces, diametrically opposite from each
other, to stop the vehicle. The symmetric braking force can avoid any large bending
moments, air gap-closing forces and a lighter rotor results. Moreover, the size of
wheel rim has been limited at least 18 to accept the motor but it is also important
that the offset of the mounting flange allows the standard vehicle track and therefore
steering geometry to be maintained (Whitehead, 2012). A simplified vehicle model
for the architecture of the in-wheel motors is built by using the existing suspension
system adds a dummy part on the wheel centre to simulate the in-wheel motors
acting on the un-sprung mass as shown in Figure 5-3.
Figure 5-3 Architecture of in-wheel motors model
5.3 Modelling and validating of in-wheel motors vehicle
The Ride Rig model used to test the dynamic behaviour of the vehicle model
described was in the previous Chapter. This particular model has 10 degree of
freedoms in total: the vehicle body has six degree of freedoms - three in rotational
and three translational, and the suspensions each have one degree of freedom. The
86
tyre is represented by a single spring damper which has the same damping
coefficient and vertical stiffness as the tyre.
Figure 5-4 Westfield I-racer with in-wheel motors in ADAMS
Figure 5-4 shows the model built to represent the layout with in-wheel motors. In this
model, the two motors connect from the middle of the rear axle to the individual rear
wheel hubs. Thus the un-sprung mass is increased by 30 kg on the rear left and right
sides. The simplified transmission system as modelled omits the driving shafts and
constant velocity joint on both sides compared to the project vehicle model. Instead,
the two electric motors are directly fitted at the centre of the wheel knuckles. It is
good practice to examine the model for its eigenvalues to find the natural
frequencies of the body and the suspensions. The model is assembled into its Ride
Rig configuration and the motion of the input jacks is set to zero. The approach used
for the calculations in ADAMS is the so-called small perturbation method in which a
small disturbance is introduced at each degree of freedom in order to construct the
stiffness and mass matrices, which are then solved in the traditional manner.
Figure 5-5 shows a complete model used for handling simulation which includes the
vehicle body, the suspension systems with a Pacejka tyre model and a closed-loop
driver model. This is closer to the project vehicle model described in the previous
chapter. Only the position of the electric motors is different. In order to check the
87
drivability and the stability for the model with in-wheel motors during the manoeuvres,
the modification in the driveline is to move the application points for the driving
torque from the middle of the axle to the wheel centre as shown in Figure 5-5. In this
analysis, there is only an identical driving torque applying to each wheel that may be
continuously produced even when the grip is lost between the road surface and the
tyre. This means the model can progress beyond the point of vehicle spin out during
a cornering manoeuvre.
Figure 5-5 Model definition for in-wheel motors
5.4 Modelling architecture of in-line motors vehicle
The model for the vehicle with In-line motors is based on the basic model which is
described in the previous Chapter. The basic model used for a kinematic check only
has the suspensions and the vehicle body, where the vehicle body is clamped to the
ground and the individual wheel centres have vertical stroke applied while the
dampers and the tyres are absent. Modelling the vehicle with the in-line motors
involves attaching the two electric motors and the drive-shaft to the rear axle. The
inertia properties and mass for the additional parts are the same as the real
components. More details were introduced and the model geometry is shown in
Figure 5-5. In fact the architecture of the in-line motors vehicle model includes the
characteristics of the conventional suspension. This means any modifications can be
omitted thereby the costs of redesign and development will be reduced but the
stability and durability is the same as the previous version.
88
Figure 5-4 the model geometry of in-line motors for kinematic test
It is necessary to build a new model for the dynamic check because the driveline for
the In-line motors vehicle was redesigned. The additional parts may change the
structural stiffness and mass matrices. This can cause a small disturbance at each
degree of freedom. Note that redesigning the driveline may have some flexibility to
prevent substantial constraints between the multiple rigid bodies. Thus the driveline
allows the drive-shaft to have some movement along the axle during ride height
change. The weight distribution for the project vehicle is 50/50 and the additional
masses from the electric motors can be balanced by mounting battery packages in
the front of vehicle. Figure 5-6 shows the model still has 10 degree of freedoms due
to the motors and the drive-shafts being attached to the vehicle body and rotating
with the wheels. Thus no extra rotational and translational movement is required,
again with the dampers and tyres absent.
Figure 5-6 Dynamic check model for in-line motor
89
Figure 5-7 shows a completed model for a drivability check in which the dampers
and tyres are now present. According to the configuration for the In-line motors
vehicle model, the points for applying the driving torque move to the position where
the motor and the drive shaft connect together. These retrofits to the architecture are
intended to improve vehicle drive efficiency in which the new driveline has less
mechanical components and a direct drive design. The efficiency of a conventional
driveline is about 20% and an electric vehicle with in-wheel motors increases to
approximately 80% depending on the driving cycle. However, the common
architecture for in-wheel motors vehicles includes a shorter driveline than in-line
motors vehicles, but requires major modifications to traditional suspensions to
include the power source.
Figure 5-7 A full vehicle model for drivability check
90
5.5 Simulation Results
The results are carried out by simulating the ADAMS models in several examinations.
Primary concerns with the additional mass on hub wheel centre are degraded road-
holding and ride comfort. It is unacceptable to use a single measure in evaluating
vehicle performance. Instead, there are a number of different indicators available to
be considered as good practice. These indicators can be represented by using
objective results, for example, numerical measures and data from predictive
modelling or subjective review by an expert driver.
The performance of the ground vehicle dynamics can be broadly split into: ride,
refinement, safety and driveability. The ride comfort is the capability of the vehicle to
isolate disturbances and the refinement is the capability of the vehicle to reduce
noise and vibration. The safety is the capability to drive and stop in emergency
situations, and the last aspect is the driveability that is the agility of control response
of the vehicle, for example, steering and braking in normal situations. The
simulations focus on objective measurements and subjective analysis to review the
influences on vehicle dynamic behaviours with additional un-sprung mass. Some
subjective assessments will be reviewed to prove the results.
5.3.1 Analysis of ride comfort
Before starting analysis the simulation results, it is worth to review the function of
subjective assessment that has been used in vehicle industry over decades.
Subjective methods were first developed in the aircraft industry, where the Cooper
rating scale was used to rate aircraft in terms of the ease or difficulty in completing
specific tasks. A similar method used in vehicle industry is call Vehicle Evaluation
Rating which the scale goes from 1 to 10- 1 is worst and 10is best. In practice, if a
vehicle has been marked by grades 1 to 5, that means the vehicle is unable to sale
on the market and unsuitable for further testing. A middle range of grades, such as
scores 6 to 8, are used to denote vehicles that acceptable, mid-class and excellent.
There are only skilled practitioners are available within the industry to acquire the
subjective review of vehicles.
91
According to Anderson and Harty (2010) have tested a 2007 Model Year Ford Focus
that was added 30kg additional mass to each wheel centre. This approach assumes
rotating and non-rotating un-sprung masses in a way which broadly reflects the in-
wheel motors. No other changes were made to the vehicle, which is to say no
development was performed for the purpose of this exercise.
Subjective Figure 5-8 shows the subjective result plot for ride evaluation. Some
degradation can be found in the pitch control, small impact feel and large impact feel.
Nothing in need of attention in the roll movement is described by the skilled
practitioner.
Figure 5-8 Subjective results for ride testing (Anderson and Harty 2010)
Objective As well as the subjective reviews carried out, objective measurements

are made of ride comfort by using the in-line and in-wheel motors models that were
built in previous section. The vehicle body is free and the platforms under wheels
impart surface motion into the suspensions. The tyres represented by single spring
damper and can separate from platforms under large inputs. The simulation runs at
the dynamic mode for two second and to examine the vehicle models for their
eigenvalues Natural Frequencies at each degree of freedoms. The results show in
Table 5-1. The information includes: sequential number of the mode that was
predicted by the Eigen solution, natural frequency corresponding to the mode,
damping ratio for the mode and list the real and imaginary part of the eigenvalue.
92
It is often true with passenger vehicles that the primary ride roll mode is at a higher
frequency than the other two primary ride modes, since the anti-roll (stabilizer) bars
add a large amount of stiffness to the roll mode. Wheel hop frequencies are high
compared to passenger car values but again the racing nature of the vehicle means
the vertical tyre stiffness is relatively high, increasing up the wheel hop frequency.
The four wheel hop modes are sensibly positioned with respect to each other, with
the front modes, influenced by the steering gear mass.
Table 5-1 Display the eigenvalues in tabular form
Table 5-2 Explanation of natural frequencies at each mode
Mode Frequency Comment

1-3 <0.2Hz Body on rig stabilisation springs fore-aft, yaw ,lateral
4 1.389 Hz Primary Ride 1 Heave
5 2.465 Hz Primary Ride 2 Pitch
6 3.487 Hz Primary Ride 3 Roll
7 12.56 Hz Rear left wheel hop
8 12.68 Hz Rear right wheel hop
9 18.94 Hz Front left wheel hop

10 19.14 Hz Front right wheel hop
93
The calculation for eigenvalues and eigenvectors can be performed using a function
of ADAMS which provides a prediction of the models natural frequencies and mode
shapes. After the calculation is complete, ADAMS can display the Eigenvalues in
tabular form and also plot complex Eigenvalues scatter as shown in Figure 5-9.
Finding the stability of the system depends on the real and imaginary parts of the
Eigenvalues, along with the positive or negative signs of the real parts and the
difference of their values.
The system is unstable when the sign of the real part is positive, and its behaviour is
more like an unstable oscillator. Comparatively, the system is stable when the real
part is negative. To consider a more complex system, it is important to note that the
system should has all real parts of the Eigenvalues with the negative values, and it is
necessary condition to determine the system is stable. In additional, when the
system is stable, the Eigenvalues should have the imaginary parts and that are not
equal to zero. However, the model is a stable system in which the all real values are
negative and the system has imaginary part that a pair of the values corresponds to
the each mode.
Figure 5-9 Eigenvalues for in wheel motors model
94
Animating the modes is able to give a better understanding of the deformations for
the vehicle body and suspensions as shown in Figure 5-10. For example, the first
animation is the body movement in fore-aft and then the second and third animations
show the yaw and lateral movement. These three modes stay in very low
frequencies. Start from the mode 4 is the animations for Primary Ride of the vehicle
body in heave, pitch and roll, that can be seen the frequencies are increased at
these modes and these changes should effect on the performance of the ride
comfort as it is known in the vehicle dynamics field.
(a) Mode 1 Fore-aft (at 0.049185Hz) (b) Mode 2 Yaw (at 0.067399Hz)
(c) Mode 3 Lateral (at 0.15296Hz) (d) Mode 4 Heave (at 1.3989Hz)
95
(e) Mode 5 Pitch (at 2.46543Hz) (f) Mode 6 Roll (at 3.48793Hz)
Figure 5-10 The animations for vehicle body
It is often true with racing vehicles that the roll mode of primary ride is at a higher
frequency than other two primary ride modes, since the anti-roll bars add a large
amount of stiffness to the roll mode. In addition, the vehicle body as modelled is built
by using a single rigid body that may give large torsion stiffness for the vehicle
structure in roll mode. As well as known, the stiffness of body structure is another
open task to improve the vehicle handling and drive performance, likewise effects on
safety requirements as a major issue in vehicle industry. Figure 5-11 shows the
animations in wheel hop modes at front and rear suspensions.
(a)Mode 7 Left wheel hop (at 12.563 Hz) (b) Mode 8 Right wheel hop (at 12.681Hz)
96
(c)Mode 9 Left wheel hop (at 18.947Hz) ( d) Mode 10 Right wheel hop (at 19.143Hz)
Figure 5-11 Front wheel hop modes for in-wheel motors
The same procedure is used to simulate the in-line motors model in the ride comfort
method. The table of the Eigenvalues and the plotting of the eigenvalues scatter are
also carried out that the results are able to check how the frequencies at each mode
effect on the ride performance and the model with new driveline is stable or unstable
as shown in Table 5-3 and Figure 5-12. In tabular form, the modes for the in-wheel
motors model have been checked by using the animations in which they have same
the sequence as the in-wheel motors model, thus they can easily be compared.
Figure 5-12 The plotting of Eigenvalues scatter for in-line motors model
97
Table 5-3 Natural Frequencies for in-line motors model
Comparing the results of the objective measurement at the ride comfort check are
carried out as shown in Figure 5-11 and the differences in the natural frequencies
can be found between the two different architectures. For the in-line motors model,
the natural frequencies at the mode 2 and mode 3 (i.e. the body motion in yaw and
lateral direction) have very small variety comparing to the in-wheel motors model.
The change is because of adding the in-line motors driveline effects on the stiffness
and mass matrices of the vehicle body.
For the in-wheel motors model, the extra masses are attached on the suspensions
merely, thus no influence on the natural frequency of the vehicle body. In addition,
more decrease in the natural frequency occurs at the mode 7 and 8 (i.e. the wheel
hop modes for the rear suspensions). As previously note, the reason for frequency
reduction is due to the un-sprung mass being increased by integrating all the
components into the wheel hub. However, the wheel-hop mode of vibration is
reduced in frequency from around 17 Hz on the in-line motors model to around 12
Hz with in-wheel motors.
98
It is clear shown in Figure 5-11 (a), which the frequencies at the rear wheel hop
modes are obviously different, and that should change the level of response, but not
significant. There are no testing results to indicate that the vibrations of unsprung
mass from 12 to 17Hz will produce more noisome, thus it may be summarized that
the ride behaviour is not substantially changed by adding the masses on the wheel
centre.
Table 5-4 Comparison of natural frequencies for in-line and in-wheel motors
model
(a) In-wheel motors (b) In -line motors
Nevertheless it is necessary to introduce another simulation for further ride comfort

check in which an inspection in the vertical wheel hub acceleration will be needed to
verify any influences of ride behaviour due to the additional masses. A typical
method has developed by Blundell (2004) who defined a motion imparted to a jack
part to represent inputs from the road surface, and a step function can also be used
to describe a profile of speed bump as shown in Figure 5-13. Motion in translational
applies on the jack to give the suspension an severe impact, and the result shows in
Figure 5-14 over a large single disturbance the in-wheel motors model gives
measurably poorer behaviour higher and severe response, that can elucidate why
the in-wheel motors vehicle provides lower score on subjective assessment.
99
Figure 5-5 Road profile for speed bump (Blundell, 2004)
Figure 5-6 Measured results for wheel hub acceleration
5.3.2 Analysis of drivability check
As mentioned, it is very important to work out how the additional masses will effect
on the drive safety and control performance of electric vehicles with different
architectures. Reasonably, some reviews in subjective evaluations of real vehicle
testing environment will be carried out as following, and therefore the driver
sensitivity at the level of expert can define the response of the vehicle precisely and
100
give an authentic score in vehicle evaluation rating. The previous comparisons
mainly focus on the ride comfort check with extra masses on the rear wheels, and
that are very useful evidences to elucidate the advantages of architecture for in-line
motors vehicle. More objective measurements in drivability will be accomplished still
with more interested results.
Subjective The subjective result of drivability check is shown in Figure 5-15, and
the largest deficit concern is the effort in the steering in which becomes heavier
under a large range of circumstances. The degradations obviously effect on the
drivability of the in-wheel motors vehicle at obstacle avoidance and emergency turns,
even need more efforts on parking. Nevertheless, the architecture for in-line motors
vehicle has a typical or traditional chassis in which the drivability can be expected as
same as a performed vehicle.
Figure 5-15 Subjective results of steering behaviour (Anderson and Hart, 2010)
Objective The previous models for the drivability check are taken into the
computer-based (ADAMS) simulation. The simulation consists of two manoeuvres:
steady state cornering and double lane change. For first step, the analysis for steady
state cornering is carried on a 40m radius course, and a closed-loop driver model
uses PID controller in which strong emphasis on Integral gain gives the model
accurate path following. The simulation starts with a speed of around 5kmh and the
101
driveline delivers equal torque to each wheel attempting to track a speed target.
Speed ramps up at around 1kmh per second, substantially similar to the rate at
which speed is applied in a normal constant radius test.
In-line
Motors
In-wheel
Motors
Figure 5-16 Ground plane velocity during steady state cornering
The analysis ends when the vehicle can no longer follow the line, either through tyre
saturation or when the two inside wheels lift clear of the ground with an incipient
rollover event. In the Figure 5-16, the both models tends to follow the target speed
at the begin of the simulation, and then the model with in-wheel motors starts
skidding due to the tyre saturation after simulation time16seconds, at 21m/s ground
plane velocity. The model with in-line motors skids two second after this, at 22.5m/s.
The validity of the Pacejka 89 equations collapse at high inclination angles and so
the continued solution is of no interest.
102
Figure 5-17 The trajectories at cornering
More effects can be seen in the trajectories at steady state cornering as shown in
Figure 5-17. In side pushing effect of vehicle scenario, the model with in-wheel
motors reaches the performance limits quicker and exhibits more understeer
behaviour. The model with in-line motors induces extra yaw moment, helping the
vehicle reduces understeer behaviour, thereby reducing the steering angle required.
Figure 5-18 clearly shows that the centripetal acceleration is less with in-line motors.
103
In-wheel
Motors
In-line
Motors
Figure 5-7 Testing results for centripetal acceleration of in-line and in-wheel motors
In the next step, the models will be taken into double lane change manoeuvre. The
manoeuvre is essential a scaled version of ISO3888 lane change in its geometry,
only the test protocol differs slightly. It consists of a defined zone in which the vehicle
can take any path and reflects a real world avoidance manoeuvre within a finite width
road. The vehicle must displace laterally by around 3.5m minimum and then return to
its original path. The modification of the double lane change is intended to reflect
maximum effort manoeuvring at highway speeds and the length and width of the
manoeuvring zone are related to the vehicle proportions, as shown in Figure 5-19.
Also the details for optimizations of double lane change will be discussed in the next
Chapter.
104
Figure 5-8 Double Lane Change Test Course
With real vehicle, the driver skill levels modify the performance of a given vehicle
through the lane change substantially and so when comparing vehicles it is
preferable to use a panel of drivers or at least a consistent driver. In the analysis the
driver model is consistent and repeatable as well the closed-loop PID control gives
the vehicle an accurate path following.
For the simulation, the both models are driven up to the manoeuvre start at 100km/h
and the throttle released. The overrun condition is typically the most difficult in terms
of stability. Although the vehicle loses speed as it enters the second of the two
manoeuvres, it is typically unsettled dynamically and so the second transition is often
the more problematic of the two. The maximum variation of yaw rate can be seen in
the stage of centre correction and exit oscillations, there is that the increased angular
velocity in yaw around 8 degree/s emerges an overshot at the peak values
comparing to in-line motors model as shown in Figure 5-20. Also some influences of
body slip angle are shown in Figure 5-21, which is important because it effectively
grows the width of the vehicle. The changes are large enough to indicate the driver
the vehicle will be lost control. The measurements based on computing simulations
are broadly familiar with subjective assessments, which more evidences will be
shown in next section to see somewhat worse results as expected.
105
Exit oscillations
Centre correction
Figure 5-9 Comparison of yaw rate for in-line and in-wheel motors in DLC
Figure 5-10 Comparison body slip angle for in-line and in-wheel motors in DLC
Considering driver steering motion inputs for both models, it can be seen in Figure 5-
22, the model with the in-line motors has a large reduction about 10 degrees in first
reversal and centre correction section, also a useful improvement in overall
magnitude of steering angle inputs can be found. In addition, there is an improved
quality to the steering traces in terms of smoothness and lack of reversals when the
model with in-line motors. It may expect more significant differences in driveability
106
check when comparing the two architectures, nevertheless considering the change
on in-wheel motor model is only to add 5% of the entire vehicle weight into the rear
wheel centre and the extra weight is even less than a passenger.
First reversal
Exit oscillations
First turn in
Centre correction
Figure 5-11 Comparison of steering angle for in-line and in-wheel motors in DLC
5.5 Conclusion
In this chapter, the vehicle architectures of in-wheel motor and in-line motor are
introduced that the significant modifications of original suspension have been done
to integrate all components of in-wheel motors, such as braking system, cooling
system and electric motors, into the constraint volume at inside of the rim.
Furthermore, the un-sprung mass is obvious increased for implementing in-wheel
motors on a vehicle, and degraded ride and handling performance. Compare to the
architecture of in-wheel motors vehicle, the in-line motors vehicle were required the
minimum variations of suspension system thus is can remain the un-sprung masses
and reduce redesign costs.
Objective measurements were carried out by using several computing simulation

models in ADAMS, such as kinematic, dynamic and handling models. The aspects of
simulation have been examined in detail and can be concluded thus the overall
performance of ride comfort is demoted and the obvious impact of steering
107
behaviour emerge with testing the vehicle models in cornering and lane change
manoeuvres. The effects on safety and driveability are required some variations in
suspension component detail and increase in damping levels to restore agility.
Overall, developing the electric vehicle in next recent state the architecture of in-line
motors vehicle can preserve the characteristics of conventional vehicle in high
performance, and the redesign costs and efforts can be economized on improving
capacity of electric motors and battery life. Moreover the good potential of individual
wheel motor control can be implemented by using in-line wheel motors for
substantial improvements in vehicle behaviour.
108
6 Torque vectoring system with in-line motors
6.1 Introduction
Active intelligent control systems for achieving vehicle stability and handling have
been developed and implemented to enhance the driving safety and performance of
the driving vehicle. Some enhancements, such as Active Steering Systems and
Electronic Stability Programs, can help the driver to retain control of their vehicles
when the grip between road surface and tyre is lost. In previous investigations, the
ABS based Stability Control Systems are principal safety implement to accomplish
the safety requirements under adverse road conditions. However, the vehicle speed
is degraded while the Stability Control System implements braking force on four
wheels individually to improve the correct position of the vehicle body. Moreover, the
Torque Vectoring (TV) system can be designed to improve the vehicle handling
qualities and avoid the vehicle speed decrease, or in other words the fun-to-drive
aspect. Thus, torque vectoring can be used to influence the driver experience.
In this chapter, the investigation focuses on developing and designing a novel TV

control strategy that is called Optimal Driveline Control Strategy (ODCS). The ODCS
involves three levels of control: Desired Dynamics Behaviours, Secondary Control
and Advanced Torque Vectoring Control, more details of which will be represented in
the following section. Reviewing the basic principles for TV control on the
conventional driveline helps to understand how to implement those control strategies
on the pure electric vehicle. The configuration of the ODCS algorithm is clearly
shown in this chapter, and each level of control is explained in detail.
The project vehicle model with three different types of drivelines is taken into
Computer-Based simulations. The results are carried out by running the vehicle
model through the steady state cornering and lane change manoeuvres. The results
for the vehicle model with ODCS are also compared against the conventional
drivelines, such as Open and Limited Slip Differential, during which the vehicle
speed is increased.
109
6.2 Basic principles
The conventional drivelines, such as Open Differential and Limited-Slip Differential,

have been broadly used in the rear driving vehicles. The Open Differential allows the
driving wheels to rotate at different speeds while giving the same driving torques on
both wheels. The Limited-Slip Differential is designed to improve handling and
stability while the vehicle turns into a corner, but a fixed kinematic relationship can
only transfer the driving torques from the faster spinning wheel to the slower wheel
(Mohan and Sharma, 2006). Thus, developing a Torque Vectoring system is required:
first to retain the abilities of the open differential and limited slip differential, and then
to incorporate a means to vary the kinematic ratio across the differential thus
affecting the torque distribution between the wheels.
Figure 6-1 Definition of torque vectoring differential
(6.2.1)
(6.2.2)
(6.2.3)
(6.2.4)
110
Figure 6-1 shows a basic principle of TV applied to the rear axle when turning the
vehicle left, the engaging clutches can transmit the engine torque via the pro-shaft
from left driving wheel to right side with transferring torque . In addition, adding or
subtracting the driving torques on the left and the right wheels give a difference
on the longitudinal driving force. The difference generates additional yaw moment
to help the vehicle turn into the corner more easily.
Here is a good example in Figure 6.2.2 to show how to accomplish the requirements
of TV by using a mechanism TV differential. The system includes planetary gears
and slipping multi-plate wet clutches, which those components are controlled by
electromechanical or electrohydraulic control systems (Li and Wu, 2011). In addition,
the functions of TV system can be applied to all types of driving vehicles, such as
Rear-Wheel-Drive (RWD), Front-Wheel-Drive (FWD), and Four-Wheel-Drive (4WD),
which the device enables to be mounted on either front, rear or both axles to modify
engine torque to each wheel. Closed-Loop Control is used in the system, thus the
vehicle speed, wheel slip ratio, yaw rate error and other parameters have to be
measured. The outputs from the measurements are used to modify the driving
torques from an Internal Combustion engine.
This item has been removed due to third party copyright. The unabridged version of
the thesis can be viewed at the Lanchester Library, Coventry University.
Figure 6-2 Schematic of a mechanism torque-vectoring differential
(Mohan and Sharma, 2006)
111
Reviewing the schematic of a mechanism TV differential in Figure 6-2, the device
uses stepped planets and sun gears meshing each other to provide for a differential
action. The biasing gear-set is connected to the wet clutch as brakes that can slow
down one side and speed up on the other side thereby allowing the driven vehicle to
be turned. The secondary gears enable the reduction of the brake torque required to
steer the vehicle, and thus increase the efficiency of the system.
Figure 6-3 Explanations of the left-and-right torque vectoring
In order to explain how the functions of a TV system can be implemented on the

project vehicle clearly, a schematic that shows the relationship between driving
forces and maximum cornering forces is carried out in Figure 6-3. The project vehicle
is pure electric vehicle with the Rear-Wheel-Drive, thus a mechanism driveline is
replaced by a pair of electric motors at the middle of the rear axle where a differential
is located. Moreover, the characteristics defined in the previous chapter indicate the
project vehicle is under-steer and the weight distribution is 50/50 so the maximum
friction circles are symmetric at the front and rear wheels when driving the vehicle in
a straight line. According to the configurations of the project vehicle, only a Left-and-
Right TV control has been considered in this research. The schematic shows that
assuming the vehicle model is turning left, lateral acceleration causes the right wheel
maximum friction circle to increase and the left wheel maximum friction circle to
112
decrease. If using a function of Open Differential on project vehicle, for example, the
driving torques are identical at two rear wheels and only the right tyre generates
cornering force because the maximum tyre friction force is bigger than the
driving force. However, using a Left-and-Right TV configuration is possible to
redistribute the engine torque between the two wheels, thus the left driving force will
be and the right driving force is , so that both the driving wheels
generate the maximum cornering forces compared to the vehicle without TV control.
The additional cornering force can be expressed as below:
(6.2.5)
So ( ) ( ) ( ) (6.2.6)
: Left and right maximum cornering force with TV
: Right maximum cornering force without TV

: Left and right maximum tyre friction force
: Driving force
Based on this formula, when the vectoring torque is applied, the additional
cornering force will become a maximum value as shown below. It shows that the
vectoring torque generates the additional cornering force.
(6.2.7)
In order to implement this concept into the handling and stability control of the project
vehicle, a control strategy will be designed in the next section. Moreover, the
promise of individual motor control shows good potential for substantial
improvements in vehicle behaviour. It is worth mentioning, a mechanism TV
differential is difficult to apply the vectoring torques to the driving wheels when the
vehicle is on throttle off position. A solution has been found by adding electric
equipment to the driving axle, thus enabling the extra propulsion forces to be
113
obtained. However, those issues will also be taken into account for creating a novel
control strategy.
6.3 Optimal Driveline Control Strategy (ODCS)
An Optimal Driveline Control Strategy (ODCS) is designed and developed in this

section. The aims of the control are to improve the project vehicle handling and
lateral stability; thus, to achieve those requirements the design concepts will be
based on the Left-and-Right TV control configuration and have three levels of control.
The outputs from the control strategy can dominate the pair of electric motors
individually.
A detailed diagram of the Optimal Driveline Control Strategy is exhibited in Figure 6-

4, which consists of three levels: Desired Dynamics Behaviours, Secondary Control
and Advanced Torque-Vectoring Control. The first level is created to define the
operating scope and desired behaviours. The desired behaviours are able to be
found according to the vehicle speed and path curvature calculation. The secondary
control is created to implement the desired driving torques and yaw rate error
correction for tracking the desired behaviours. The inputs from the upper levels
should dominate the electric motors through the Advanced Torque Vectoring Control
(ATVC). The use of ATVC level is to modify the vectoring torques and driving
torques outputs, and some constraints are added to avoid wheel slip while
accelerating or braking.
114
Figure 6-4 Schematic of Driveline Control Strategy (DCS)
In addition, a human driver model is required to simulate the project vehicle. The
performance of a real vehicle, through road testing, can be modified by driver skill
levels; so it is preferable to use a consistent driver when comparing the vehicles.
Moreover, the aim of the DCS is to assist the driver to cope with the extreme driving
scenarios even in the throttle off condition. However, an Open-Loop Driver model is
used with the project vehicle to follow a desired path with consistent driving
115
performance. Thus the simulated results from the model with different driving
configurations and vehicle speeds can be compared.
6.3.1 Definitions of Desired Dynamics Behaviour
Reference model definition
The primary requirement of the ODCS is to deliver desired dynamics behaviour in

terms of the desired path, desired yaw rate, operating scope and vehicle linearity.
The vehicle as modelled in the previous chapter will be taken into a steady state
cornering manoeuvre for the first simulation. The yaw rate transfer function of the
vehicle without TV control (the so-called baseline vehicle) can be obtained from the
equations describing the lateral force and yaw dynamics of the single-track vehicle
model (Milliken, 1995). Thus

(6.2.8)

In this equation, is the mean steer angle of the front wheels. The stability
derivatives can be expressed as a function of the front and rear cornering stiffnesses,
i.e. and , respectively, as
These equations are correctly referred to as a 2-degree-of-freedom model; they are

sometimes referred to as a bicycle model but this description should not imply that
the description may be suitable for two-wheeled vehicles.
The next hurdle to be crossed is the representation of the intended behaviour of the
vehicle the reference states. Competition-developed lap simulation tools use a
track map based on distance travelled and path curvature. This representation
allows the reference path to be of any form at all and allows for circular or crossing
116
paths to be represented without the one-to-many mapping difficulties that would be
encountered with any sort of y-versus-x mapping. Integrating the longitudinal velocity
for the vehicle gives a distance-travelled measure that shows it to be tolerably robust
against drifting within simulation models. Using this measure, the path curvature can
be surveyed in the vicinity of the model.
Some researchers favour the use of a preview distance for controlling the path of the
vehicle, with an error based on the lateral deviation from the intended path. However,
there is usually a difficulty associated with this since the lateral direction must be
defined with respect to the vehicle. Failure to anchor the reference frame to the
vehicle means that portions of the path approaching 90 degrees to the original
direction of travel rapidly diverge to large errors. Projecting a preview line forward of
the mass centre and located on the vehicle centre line is unsatisfactory due to the
body slip angle variations. Either the proportional gain must be reduced to avoid pilot
induced oscillation type behaviour, which leads to unsatisfactory behaviour through
aggressive avoidance manoeuvres, or else some form of gain scheduling must be
applied.
An alternative method, used here with some success for a variety of extreme
manoeuvres, is to focus on the behaviour of the front axle. This model fits with the
experience of drivers at or near the handling limit, particularly on surfaces such as
snow where large body slip angles highlight the mechanisms used in the drivers
mind. High performance driving coaches (Palmer, 1999) rightly concentrate on the
use of a model the driver needs in order to retain control in what would otherwise
become stressful circumstances of non-linear vehicle behaviour and multiple
requirements for control typically vehicle orientation (body slip angle) and velocity
(path control). Useful learning occurs on low grip environments that can be readily
transferred across to high grip. In low grip environments, the extreme non-linearity of
response of the vehicle can be explored at low speeds and with low stress levels,
allowing the driver to piece together a model to be used within their own heads; it is
then a matter of practise to transfer the lessons to a high grip environment. The
same concepts can be used to explore the behaviour of a driver model within
MSC.ADAMS.
117
Desired Yaw Rate
This section describes the methodology for the definition of desired yaw rate for the
controller implemented on the vehicle model with nonlinear dynamics. A target
under-steer characteristic is defined in terms of an analytical function relating the
dynamic steering-wheel angle (where is the actual steering-wheel
angle, and is the kinematic steering-wheel angle) to the lateral acceleration .
Therefore, Shibahata (1993) proposed a method based on the following three
characteristic parameters: the under-steer gradient ; the threshold
value , which defines the upper limit of the linear part of the under-steer
characteristic; and the maximum lateral acceleration achievable in trimmed
conditions . Thus
( )
( )
Where ), and . The terms , , and

can be chosen according to the control design requirements. In particular, the same
value of at different values of the longitudinal acceleration have been considered,
to achieve the compensation of the variation of the under-steer gradient in traction
and braking conditions. Following Shibahata (1993), the steady-state value of the
desired yaw rate is given by
All equations allow the generation of the desired yaw rate as a function of vehicle
speed, steering-wheel angle, longitudinal acceleration, and friction coefficient at the
tyre-road contact. At the starting point, finding a desired path for testing the vehicle
model under dynamics manoeuvres, i.e. steady state cornering and double lane
change, is important and it can be defined by using path curvature . The analysis
starts when the vehicle runs on a 33m diameter circle, and then a desired yaw rate
can be calculated by using the expression above.
118
The path curvature for a circle should be a constant value, but it needs some
transition curvatures with varied radius to link path curvature from the straight line to
the constant radius circle. The approach can avoid abrupt steering inputs while the
vehicle model traces the desired path. The desired path curvature can be converted
to a SPLINE command as used in ADAMS, as shown in Figure 6-5. The X values
are path lengths depending on vehicle speed, and Y values are path curvatures. A
human driver model as motioned in Chapter 4 will apply a torque on the steering
wheel to drive the vehicular model; the inputs from the driver model are
corresponding to the path curvature (SPLINE) commends.
Figure 6-5 Path curvature spline in ADAMS for cornering
It is tempting to draw the lane change path as a series of connected arcs the
treatment of OHara, (2005) is typical - but this does not reflect the reality of a
drivers inputs, which are observed to be somewhat fluid and continuous during a
high effort lane change without the dwell periods a series of arcs would imply. An
improvement might be to view the path as a cosine ramp of the form. is the wheel
base of the project vehicle. The length of the entrance section is rescaled and the
path of cosine ramp lane change is created in Figure 6-7
( ( )) (6.3.1)
119
Lat eral Pat h ( m) 0
0 5 10 15 20 25 30
-3.5
Distance Travelled (m)
Figure 6-6 Cosine ramp lane change path visualisation
When considering the task in detail, it is noted that the mid-section is wider than the
vehicle. When the position of the vehicle is fractionally away from the very edge of
the entrance gate, it is necessary to tune the relative length of the turn-in and turn-
out sections and to run out to the very widest part of the path available. Further slight
modification to the exact form is made to make it continuously differentiable as
shown in Figure 6-8. If this optimized cosine form is compared with the initial cosine
form, a substantial reduction in centripetal acceleration is realised as shown in
Figure 6-9.
0
0 5 10 15 20 25 30 35 40
-0.5
In sid e Wh eel Lat eral Pat h ( m)
-1
-1.5
-2
-2.5
Optimised Cosine
-3 Cosine
-3.5 NATO DCL Entrance
NATO DCL Mid Section
-4
NATO DCL Outer Limit for Inside Wheel
-4.5
-5
Distance Travelled (m)
Figure 6-7 Optimized lane change path with the behaviours of real driver
120
1
0.8
Peak Lateral Acceleration
0.6
0.4
(g)
Cosine
0.2
0
20 40 60
Vehicle Speed (km/h)
Figure 6-8 Comparison of the lateral acceleration for cosine and optimised lane change path
Figure 6-9 Path curvature of lane change
By using the same calculation as the cornering manoeuvre, Figure 6-10 shows that
the path curvature for the double lane change is created and then it can be
converted to a SPLINE command. Thus, the desired yaw rate and lateral
acceleration can be found. A comparison can be made between a theoretical value
and a base vehicle simulation at 30 km/h and shows that even at this speed the
response is being limited by the dynamics of the vehicle; nevertheless there is broad
agreement as shown in Figure 6-11.
121
0.25
Ay(g)
0
ADAMS Ay(g)
-0.25
-20 0 20 40 60 80 100
Figure 6-10 Comparison of desired and actual lateral acceleration
An actual double lane change protocol calls for an increase in vehicle speed until the
driver is no longer able to complete the manoeuvre; the outcome from the test is the
highest speed at which the driver can repeatedly complete the manoeuvre. For this
simulation, the vehicle model runs from low speed 40 km/h to a high speed that is
close to the physical limitation of the vehicle model. Repeated iterations are made to
discover the optimal performance of the vehicle model that should be improved by
Driveline Control Strategy. Nevertheless, the reported vehicle behaviour is to
complete the manoeuvre at 110 km/h so that the lateral acceleration starts to exceed
the control region of the TV system. The ODCS is no longer to fulfil the safety
requirements, therefore the manoeuvre approaches an emergency stage requiring
the vehicle to reduce the speed to control its motion and diminish the kinetic energy
of the vehicle body by normal use of the braking or stability control system.
122
Figure 6-11 Yaw Rate Gain (YRG) for three different vehicles (Blundell and Harty, 2004)
Normally, the vehicles need to be designed to reduce the Yaw Rate Gain (YRG) at
high speed to ensure the drivers still control the vehicle when faced with emergency
evasive manoeuvres. For typical road vehicles, the characteristics of the vehicle are
performed to have further modifications to reduce the yaw rate gain even when the
tyres are not saturated. These modifications are very different between different
demands as shown in Figure 6-12, in which all three vehicles are under-steer in their
linear regions, but yet the racing vehicle tends to increase yaw rate gain at higher
speed thus improving dynamic control performance for expert drivers because it is
closer to neutral steer. Reviewing the YRG characteristics for the WRC 2003 car,
some linear YRG points are very close, and are even over the neutral steer line
when the vehicle speed is around 50km/h. Moreover, the linear YRG points are
substantially decreased while the vehicle speed is over 100km/h.
Based on the YRG characteristics for racing vehicles, the TV control should
dominate the YRG of the project vehicle model close to their theoretical YRG curve.
Therefore, Figure 6-13 represents an operating region for the TV system in this
research. The theoretical YRG curve for the project vehicle was calculated in the
Chapter 4. Because of the characteristics being very close to the neutral steer when
the speed is below 80km/h, the use of TV control should maintain the model in its
own characteristics and prevent a bulk over-steer. The actual YRG may have a large
123
drop as shown while the vehicle speed is raised, thus The TV control is also required
to dominate the actual YRG approaching the theoretical curve.
Neutral
Steer
Torque
Vectoring
Authority
Theoretical
YRG
Figure 6-12 Explanation of torque vectoring authority envelope
In Figure 6-13, the YRG for neutral steer and the theoretical YRG of the project
model can be calculated by using the equations below:
(6.3.5)
: Vehicle speed
: Vehicle length
: Geometric
Note that the over-steer and under-steer canyawbe
rate
easily known by a ratio of the
geometric yaw rate and wheel steer angle as shown in the equation 6.3.5. If the ratio

is more than 1, the vehicle is under-steer and if the ratio is less than 1, the vehicle is
over-steer.
124
6.3.2 Secondary Control
The aim of secondary control design is to control the vehicle model tracking the
desired dynamics, which consists of the aspects of vehicle speed control, yaw rate
correction and control modes regulator. Rolling resistance and aerodynamic drag
forces are applied on this vehicle model, thus the vehicle speed is lost and also the
vehicle loses momentum during the resistant of the tyre cornering forces. Modelling
a speed compensation controller is required to offset the vehicle speed and the
inputs for the controller are dependent on the difference in the vehicle speed that is
measured between the ground plane velocity and the speed of the vehicle model.
Another controller for yaw rate correction is designed to minimise the difference
between the desired and the actual yaw rate. The outputs from this controller have a
substantial effect on the vectoring torques control. Finally, a regulator is added into
the Secondary Control to switch the Optimal Driveline Control Strategy (ODCS)
between the driving modes.
Speed Compensation (SC) Controller
In this simulation, a constant velocity of the vehicle needs to be maintained.

Therefore, a simple method of applying torque to the driven wheels is explained in
this section. The secondary control includes a Speed Compensation (SC) Controller
as part of the traction system, which means the torques are impart to the road
wheels and then produce tractive driving forces at the tyres. The ground plane
velocity as a reference speed can be calculated by using equation 6.3.6.
(6.3.6)
Hence, the tractive driving torques from the SC controller depend on the maximum
driving torques from the pair of electric motors, and the difference between the
ground plane velocity and longitudinal velocity of the vehicle model as shown
in equation 6.3.7
( | |) (6.3.7)
125
Based on ground reference frame, the direction of the velocity is negative so
using an absolute value for the velocity is necessary to ensure the driving torques
outputs in positive, which means the positive value gives the vehicle model a tractive
torque and the negative value may cause the vehicle to slow down. Note that the
maximum torque as output from the electric motors may produce large instant
tractive forces at the driven wheels when starting the simulation. It is necessary to
modify the torque to prevent the driving wheels slipping when the simulation
starts at different initial speeds. Therefore, an optimal output from the SC controller
will be taken into account in the third level of the ODCS.
Yaw Rate Error Correction (YREC)
The path curvatures for steady state cornering and lane change were detected in the
first level of the ODCS and hence the Desired Yaw Rate is found from the Path
Curvature and forward velocity using:
(6.3.8)
To apply the equation 6.3.8 to the simulation software, it needs to find out the path
length when the vehicle model runs at various speeds. Calculating the path length is
simply to integrate the ground plane velocity as shown below.
( )
6.3.9)
The body slip angle can be computed by using the equation 6.3.10 based on the
velocities of and .
( ) 3.10)
126
Hence, the variables of lateral acceleration and longitudinal acceleration are
created and calculated in the vehicle body axis system. The units are set up in m/s 2.
As follows, the centripetal acceleration is given from the components of
acceleration and using
(6.3.11)
The front axle no-slip yaw rate is found from the components of the centripetal
acceleration , the yaw acceleration , the distance , and the ground plane
velocity using
(6.3.12)
Eventually, the yaw rate error is then found from the desired yaw rate and
the front axle no-slip yaw rate using
(6.3.13)
The yaw rate correction uses a close-loop control to monitor the vehicle body
position and minimise the yaw rate error. In general, a controller, such as the PID
controller, is required to be assembled into the close-loop control. It is broadly used
in control theory; and the advantage of the PID controller is that it can produce
continuous output, and it has no steps. It consists of three terms:
The term of P provides a proportional output to reduce the current error value
and multiplying a gain value can adjust the proportional response.
The term of I contributes to the response proportional to both the magnitude

and duration of the error. In the PID controller, the integral gives the
127
accumulated compensation from the previous correction and counts the total
of instantaneous error over time. Thus, the error can be minimized by
multiplying a gain value and that will be added to the controller output.
The derivative is the D item of the PID controller in which the derivative of the
error is calculated by determining the slope of the error over time, and the
derivative gain can adjust the rate of change.
In the ideal parallel form, the controller theory section for Yaw Rate Error Correction
(YREC) is shown below:
(6.3.14)
Driving Modes Regulator
When simulating a vehicle model in manoeuvres, the use of several driving modes
enables determining the differences in vehicle dynamics handling and driving
performance. Therefore, the project vehicle is set up for three driving modes as
known: free rolling, driving with a differential, and TV control. The free rolling mode is
when the vehicle model is coasting through the manoeuvres without any propulsion
forces and only frictional force acts between road surface and tyre. The second
driving mode is to apply the use of an open differential to the model driveline to
produce an even torque at the driving wheels. The developed TV control is the third
driving mode in which the TV control should retain the function of the Open
Differential and Limited-Slip Differential, and it also has the ability to modify the
driving torque. Thus, a regulator has been designed to switch the SC controller and
YREC controller on/off between the driving modes automatically. In addition, when
the SC controller is off, the vehicle model does not have any propulsion forces at the
driving wheels. With this status, as mentioned, it is difficult for the mechanical
driveline to have the vectoring torque while retaining the position of the vehicle.
Nevertheless, the regulator can still force the ODCS to implement its function to
128
control the motors while the vehicle is turning; thus the vehicle enables having the
continuous vectoring torques at the driving wheels. To explain the use of the
regulator clearly, Table 6-1 shows how it switches between the driving modes.
Table 6-1 The schematic of control modes
Control Modes Driving Modes Speed Control Yaw Rate Error

Correction
1 Coasting Deactivate Deactivate
2 Open Diff Activate Deactivate
3 Torque Vectoring Deactivate/Activate Activate
6.3.3 Advanced Torque Vectoring Control (ATVC)
In the previous section, the desired tractive driving forces from the speed
compensation and the desired yaw rate are determined. The inputs from the
secondary control level should be applied to the vehicle model through an Advanced
Torque Vectoring Control (ATVC) level. The ATVC is designed for the last level of
the ODCS and in order to modify the inputs from the previous control level. The aim
of this level should satisfy the conditions below.
Modifying the traction force for speed compensation

Optimizing the vectoring torque for yaw rate correction
Inspecting the driving efficiency
Modifying the traction force for speed compensation
In general, a speed compensation controller is derived to minimize the difference

between the ground plane speed and actual speed. The difference in the speed can
be modified by multiplying a constant gain . Normally the gain value is simply
set up for the maximum driving torque from the electronic motors; nevertheless, this
129
simplicity of the adjustment can cause some issues, such as longitudinal wheel slip,
when the vehicle runs at different speeds.
As a complex method, optimization can help and improve the overall design of the
product. If the process has a good design in early stage, the design cycle can be
shorted. In general, an optimization problem is described as a problem to minimize
or maximize an objective function over a selection of design variables, while
satisfying various constraints on the design and state variables of the system. The
objective function is normally represented by a numerical form of the quality, or
stability of the model. The aim for modifying the traction force for speed
compensation decides that the optimization chooses to find the maximum torque of
the traction as shown in Equation 6.3.15. The optimal value of this formula
corresponds to the best design possible that can give the optimal driving torques for
the speed compensation. Design variables can be thought of as the Ground Plane
Velocity for this design problem. The Ground Plane Velocity can be altered to define
the design results. In this study, the changes in the vehicle speed should result in
changes to the driving torque.
( ) | |
(6.3.15)
: Optimal torque for speed compensation
: Ground plane velocity
The difference in the vehicle speed can be minimized by multiplying an optimum gain
value as shown in Equation 6.3.16. Constraints are boundaries that directly or
indirectly eliminate unacceptable issues, thus a constraint for wheel slip is created to
keep the wheel slip ratio at driving wheels below a limit value . The constraint
only is applied when the actual wheel slip ratio is over the limitation; it also can
be explained by using the formula as below:
130
{ } | |
(6.3.16)
However, the ATVC is able to produce the optimized driving torque under the
constraint to compensate the speed. In this calculation, the design variable is set up
for the Ground Plane Velocity , thus the amplitude of the optimized torque
depends on two variables: the Ground Plane Velocity and the speed compensation.
Optimizing the vectoring torque for yaw rate error correction
A considerable change in yaw rate with vehicle speed can be perceived by the driver
as inconsistent vehicle behaviour during in normal driving conditions. To make the
vehicle behave more predictable, the TV control can be used (at least partially) to
compensate the variation of the yaw rate. However, the outputs from secondary
control are not precise enough and, thus, cannot provide a-priori definable amount of
vectoring torque.
The inputs from the yaw rate correction at the secondary control level should be
applied to the project model through the ATVC. As is known, the terms of the PID
controller can give a precise and rapid response in the yaw rate correction, but it is
still necessary to tune the gain values for the proportional, integral and derivative
terms. In general, the gain values are modified manually and are closely related to
the amplitude of the vectoring torque. Due to the variation in the vehicle speed, the
gain values over all the simulations are required in order to modify the amplitude
automatically. This means that the optimal vectoring torque inputs are able to adjust
the position of the vehicle body to reach the desired dynamic behaviours in any
conditions.
Performance Index
In this paper the parallel PID controller is used as below;
( ) (6.3.17)
131
where, and is the proportional, integral, and derivative gain, respectively. The
parallel PID controller can have complex zeros, which we have observed can result
in several peaks for the magnitude of sensitivity function in the frequency domain.
In this paper, one of the most popular ways of quantifying controller performance is
used here is the integrated absolute error (IAE)
| |
(6.3.18)
when subjecting the system to a disturbance. Both of input and output disturbances
are taken into account and chosen the weighted cost function,
( ) (6.3.19)
Where both terms are weighted equally with 0.5 to get a good balance. The and
are scaling factors from IAE-optimal PID controllers for a step load change on the
input and output, respectively.
Many methods have proven efficient and effective in special fields of application.
Such as Simplex method (Dantzing, 1963) in linear programming problems,
Conjugate gradient method (Reeves, 1964) and Quasi-Newton method (Broyden,
1967) in Non-linear Programming problems with no constraints, and Convex simplex
method (Zangwill, 1967) and Reduced gradient methods (Wolfe, 1976) in Non-linear
programming with linear constraints, etc. Researchers devote their effort to extend
these methods to wider application. The Generalized Reduced Gradient method has
been developed and proven to be one of the efficient and effective methods for the
Non-linear Programming problem with Non-linear constraints.
Based on the description of the optimization problem using the mathematical

language, the objective function (i.e. the gain values) is minimized over the selection
of the design variables. Various algorithms are available for finding a solution to an
optimization problem, thus the Generalized Reduced Gradient (GRG) algorithm that
is provided with Adams/View has been used in this study. This algorithm requires a
range of the limitation for the design variables.
132
The basic concept of GRG method entails linearizing the Non-linear objective and
constraint functions at a local solution with Taylor expansion equation. Then, the
concept of reduced gradient method is employed which divides the variable set into
two subsets of basic variables and the concept of implicit variable elimination to
express the basic variable by the non-basic variable. Finally, the constraints are
eliminated and the variable space is deduced to only non-basic variables. The
proven efficient method for non-constraints nonlinear programming problems is
involved to solve the approximated problem and then the next optimal solution for
the approximated problem should be found. The process repeats again until it fulfils
the optimal conditions.
Minimize ( )
Subject
Function and are continuous and differentiable in the

domain region }. First of all, the linearizing process at a
local feasible solution for the objective and constraint functions is performed as
follows.
(6.3.20)
Since is a feasible solution for the original problem, it must be a feasible solution
for the approximate problem, implying that
and
(6.3.21)
If is an optimal solution, the gradient of objective function must be zero, implying

that,
(6.3.22)
133
This equation is normally referred to as the reduced gradient. If the reduced gradient
at the point equal to a zero vector then it also satisfies the Lagrange stable point
conditions (Wild, 1971)
The next step is using equation (6.3.22) to calculate the gradient for the basic
variable. Consider the boundary condition for the variable and take an adequate
modification as follows

{ (6.3.23)
Checking the optimal condition, if | | then stop, otherwise, modify to construct

the searching direction . Checking the feasibility in which if at least one constraint
violates the feasibility condition, use the Newton method to pull the solution back to
the feasible region and then change the basis.
Inspecting the driving efficiency
The inspection of the driving efficiency should be taken into account in the ATVC. In
this stage, some constraints at the driving wheels should be under consideration. For
example, checking the maximum driving force during the simulation is not exceeding
the capacity of electric motors; also monitoring the driving force is not over the
maximum friction circle of the tyre at the varied vertical loads. When the speed
compensation control and the torque vectoring control operate at the same time, the
total driving torque is accumulating to approach the maximum limitation. Thus, the
function of the inspection is not only required to monitor the amount of the driving
torque, but also to check the torque at the rear wheels individually, because the
torque vectoring control may apply the different torques at each wheel.
134
Figure 6-13 Motor efficiency map
The limitation of the friction circle can be defined at each driving wheel based on the
current load and friction coefficient, which also represents the boundary of the
resultant horizontal force available at that wheel, hence the maximum friction torque
can be found as below:
| |
Furthermore, Figure 6-14 is a motor efficiency map that shows the characteristics of
the given electric motor - the torque outputs versus the motor revolution speed.
Hence, during the simulation, the maximum motor torque outputs
correspond to the revolution speed, which is restricted, based on the motor efficiency
map. Finally, the limitations can be described as below:
135
| | (6.3.25)
where, and are required driving torques at the left and right driving wheels in
which the torques are restricted by the limitations of the wheel friction circle and the
capacity of the electric motors.
6.4 Simulation results
In order to detect how successfully the ODCS can improve the vehicle dynamic
performance, the project model will be taken into two typical computer-based
simulations - steady state cornering and double lane change manoeuvres. The
project model with three driving modes has been modelled in the previous chapters.
Results are now generated to investigate the performance of the ODCS in the
project vehicle. The results are also compared against the project vehicle with the
two conventional driving modes.
The first driving mode is that the model runs only at its initial speed, which means no
driving torque is applied to the wheels, so it is called Coasting. It represents the
original driveability of the project vehicle based on its own characteristics of the
settings. The second driving mode represents an open or limited-slip differential that
is commonly used in rear drive vehicles. Finally, the third mode is the project model
with the Optimal Driveline Control Strategy (ODCS).
6.4.1 Steady state cornering
The early work of vehicle dynamics experts, such as Olley, Milliken and Segel,
developed the classical treatment of the behaviour of a vehicle and they have
documented in several textbooks dealing with the subject. As following, the classical
treatment will be summarized by a consideration based on the steady state of the
project vehicle model.
However, the treatment of the behaviour of the project vehicle is based on steady
state cornering simulation. Steady state is the condition in which, if the hand wheel
remains stationary, all the vehicle states speed, yaw rate, path curvature and so on
136
remain constant. In this simulation, a traditional evaluation method is used that
involves running the vehicle model through the defined circle at a range of constant
speeds. The variation of the vehicle speed will cause the increments in lateral
acceleration. This approach is easily performed in a practice and so it will be used
for the basis of the following discussion as shown in Figure 6-15
Figure 6-14 Vehicle model at steady state cornering
The minimum radius available in normal driving is the turning circle (Blundell and
Harty, 2004). The coefficient of friction between tyres and road determine the turning
circle and better starting point for this treatment is at low speed. Thus, the speed
compensation controller can be examined as shown in Figure 6-16, where it can be
seen a constant speed is produced during the cornering.
With speed
compensation
controller
Figure 6-15 Speed compensation with driving torque
137
There are three trajectories represented in Figure 6-17, of which the outer line is the
trajectory of the model in the coasting, and the middle trajectory is the model that
runs with the open differential. The third line is the trajectory of the model with the
ODCS that produces a function of limited-slip differential in which is closest to the
desired path.
Desired Path
Figure 6-16 Trajectory of vehicle at coasting, with driving torque and TV
The trajectory of the model in the coasting indicated that because there are no the
torques that were applied to the wheels, thus the model lost speed and exhibited
more understeer behaviours. There is a similar scenario, which has appeared on the
model with the open differential. When the identical torque was applied to the driving
wheels, it could retain the speed but there was no significant improvement on the
cornering performance, only a constant speed being given. Because the ODCS is
required to have the capacities of the open and limited-slip differential, the scenario
of the third trajectory is more like the model running with the limited-slip differential.
138
The ODCS can transfer torque along the rear axle in which the difference in the
driving torque between the left and right wheels produces more yaw moment to
make the model turn more easily, to have less under-steer behaviour, and to run with
the exact desired trajectory.
TV right
Driving force
TV left
Figure 6-17 Longitudinal driving forces for rear driving wheels
The measurements of the longitudinal driving force at the rear wheels are clearly
shown in Figure 6-18. The model in the coasting has none of the driving force, and
the open differential produces the identical driving force that is slightly different due
to the various vertical loads at the tyre. However, the ODCS can provide the required
driving force at each wheel, and the force is directly controlled by the PID controller
and adjusted by the ATVC. Also, it can be seen that the terms of the PID controller
produce the outputs value to reach the target proportionally around two seconds;
and then it makes some adjustment of the driving force outputs. The adjustment is
due to the desired trajectory needing some transition curvatures with varied radius to
link the path curvature from the straight line to the constant radius circle. After 4
seconds, the model has the constant driving force for turning the circle with the
constant radius.
139
The effects on the yaw rate correction are presented in Figure 6-19. There are four
lines for describing the yaw rates of the different driving modes. The first top line is
the desired yaw rate, and the second line just below the desired yaw rate is the yaw
rate of the model with ODCS. The third line from the top is the model with the open
differential, and the bottom line is the model in the coasting condition. Obviously, the
model with the ODCS has the best result in the cornering manoeuvre. In the ODCS,
the PID controller for the yaw rate correction can reduce the error between the
desired and the actual yaw rate, and then the integral term can accelerate the
movement of the actual yaw rate towards the desired yaw rate and eliminate the
residual steady-state error that occurs from the previous proportional term. Finally,
the derivative response improves the settling time and stability of the control. It is
worth mentioning that a high proportional gain results in a large change of the output,
i.e. if the proportional gain is too high, the vehicle can become unstable. In contrast,
a small proportional gain produces a small output response to a large input error, if
the value is too low, the effects on control may not enable the enhancement of the
dynamics performance. Thus, the function of the ATVC level can solve those issues
when the PID controller is activated.
Desired ODCS
Yaw Rate
Yaw Rate (deg/sec)
With Driving
Torque Coasting
Figure 6-18 Comparison of Yaw rate correction
140
6.4.2 Double lane change
The double lane change consists of a defined zone in which the vehicle can take any
path and it reflects a real world avoidance manoeuvre within a finite width road. The
vehicle must displace laterally by around 3.5m minimum and then return to its
original path. The length and width of the manoeuvring zone are related to the
vehicles proportions, so that the dimensions of the project vehicle with 2.69m in
length and 1.54m in width are used to define a new section for the simulation.
With real vehicles, the performance of the given vehicle through the lane change can
be substantially modified with driver skill levels and so when comparing vehicles it is
preferable to use a panel of drivers or at least a consistent driver. In this analysis, the
open-loop driver model is consistent and repeatable as it applies a steering torque to
steer the vehicle model following the desired path; although there is no guarantee it
accurately reflects the performance of real drivers. When real drivers complete the
manoeuvre, they experiment noticeably with the exact trajectory over several
attempts. This driver model is unable to learn from its previous experiences but
some modification of the intended trajectory is worth mentioning as shown in the
previous section.
The actual protocol of double lane change calls for an increase in vehicle speed until
the driver is no longer able to complete the manoeuvre; the outcome from the test
being the highest speed at which the driver can repeatedly complete the manoeuvre.
For this exercise, the model starts to run at 40km/h and the speed continually
accumulates to a higher speed of 110km/h at which point the ODCS is no longer
able to improve the dynamics behaviour. In general, when an under-steer vehicle is
taken to frictional limits where it is no longer possible to increase lateral acceleration,
the vehicle will follow a path with a radius larger than intended. So that the ODCS
can produce more turning or yaw moment on the vehicles body, more driving torque
should be applied to the outer wheel and less torque to the inner wheel. The
explanation is shown in Figure 6-20.
141
Figure 6-19 Explanation of torque vectoring control at 40km/h
The project model is driven up to the manoeuvre start and the throttle released; the
overrun condition is typically the most difficult in terms of stability. Although the
model loses speed as it enters the second of the two manoeuvres, it is typically
unsettled dynamically and so the second transition is often the more problematic of
the two. The aim of the first run is to test what the original dynamics behaviour of
the model in the coasting condition is; it is also to check how the ODCS can improve
the driveability of the model in the throttle off condition. In Figure 6-21, the results in
the yaw rate are compared, showing a large error between the yaw rate of the model
in the coasting condition and desired yaw rate; nevertheless, the ODCS is able to
minimize the error.
Desired yaw rate
With the Coasting

ODCS
Figure 6-20 Comparisons of Yaw rate without the speed controller
142
In the second run, the speed controller is switched on to give the model traction force.
Similar results can be found in Figure 6-22 in which the ODCS can still make the
improvement in the yaw rate error correction.
Figure 6-21 Yaw rate with speed controlled
In terms of stability, reducing the body slip angle is important because it effectively
produces the width of the vehicle, and because it stores energy in the pendulous
yaw mode of vibration possessed by the running vehicle. For ordinary drivers the
storage of energy in this mode is problematic to understand and control. Figure 6-23
represents a comparison of the body slip angle for the model with the different
driving mode. In the coasting mode, the vehicle body exhibits the largest body slip
angle compared to the other two modes, and the body slip angle can be reduced
somewhat when the model runs with the driving torque. It is clear to see that the
body slip angle diminishes as the model runs with the ODCS; furthermore, the clear
improvement in the driveability can be seen throughout the trace.
143
Coasting
ODCS
Figure 6-22 Body slip angle for the different driving mode
At the low speed, the use of the ODCS is more like the function of a limited-slip
differential by which means the torque can only be transferred from the fast wheel to
the slower one. When the simulation continues with an increase in the vehicle speed
to 60 km/h, it should focus attention on the yaw moment of the vehicle body. Less
yaw moment is needed, otherwise the vehicle will become over-steer, by which
means it becomes dynamically unstable with a tendency to spin out. In this case, the
ODCS is able to produce more torque on the inner wheel and less torque on the
outer wheel, and thus the yaw moment can be reduced. The explanation is shown in
Figure 6-24.
Figure 6-23 Explanation of torque vectoring control at 60km/h
144
Figure 6-25 represents the comparison of the yaw rate as the model runs with the
speed controlled. The yaw rate of the model with the identical torque overruns at the
first turn-in and exit oscillation section, and the model using the ODCS has some
improvements on those sections. There are similar results which can be found in
Figure 6-26 when the model runs at the throttle released condition. Moreover,
reduction in the body slip angle is clear evidence to prove that the ODCS enables
the enhancement of the driveability of the model, as shown in Figure 6-27.
ODCS
Figure 6-24 Variations of yaw rate with the speed controlled at 60 km/h
ODCS
Figure 6-25 Yaw rate variations at 60 km/h
145
ODCS
Figure 6-26 Variations of body slip angle at 60 km/h
The simulation is continued and the model is driven up to 100km/h and lateral
acceleration is about 0.65g. Figure 6-28 to 6-30 clearly show that there are some
improvements that can still be seen in the yaw rate and the body slip angle when
using the ODCS, but there are no substantial effects on the driveability. Due to the
high lateral acceleration, the model tends to be in accident-avoidance stage. Thus, it
needs to use the brakes to reduce the kinetic energy of the vehicle and to minimize
the wheel load variation for maximum grip and control. However, the limitation in the
lateral acceleration is the boundary for the operating region of the ODCS at high
speed.
Figure 6-27 Yaw rate variation at 100km/h
146
Figure 6-28 Yaw rate variation with the speed controlled at 100km/h
Figure 6-29 Comparison of body slip angles at 100km/h
6.5 Conclusion
Based on the basic principles of torque vectoring control, the example was carried
out to explain how the mechanical TV control improves the vehicle driveability. As is
known, re-allocating the driving torque on the rear wheel results in substantial
influence on the control of the yaw moment.
There are several promising outcomes from this work listed below; these must be
interpreted in the context of the design approach used.
147
The Optimal Driveline Control Strategy (ODCS) for RWD EVs equipped with
an in-line motor at the rear wheels and the independent motor control has
been designed to enhance vehicle driveability and yaw stability.
The ODCS involves three parts: the desired dynamics behaviours, the
secondary control and the advanced torque vectoring control (ATVC). The
use of all levels has been introduced in detail, so that the desired dynamics
behaviours have been found to define the operating region and retain the
vehicle linearity. In order to track the desired dynamics, the secondary control
has been designed that can modify the traction force then to minimize the yaw
rate error and retain the vehicle speed. Since the inputs from the secondary
control should be applied to the electric motor through the ATVC, the ATVC
has been developed to map the secondary control inputs to the electric
motors.
The simulation results with the proposed driving control have been compared
to those with the two types of other driving modes in order to verify the
performance of the proposed speed control and the proposed optimal torque
vectoring algorithm. It has been shown from simulation results that the vehicle
driveability and yaw stability can be significantly improved at the same time or
individually compared to the other driving modes.
Driving control algorithms for alternative types of electric drive system, such
as a 4WD system with independently driven in-line motors, will be
investigated in the future.
148
7. Conclusions and future work
7.1. Summary and conclusions
Overall, the thesis has shown that there are worthwhile performance advantages
available through improved torque vectoring control strategy for the project vehicle.
For example, this thesis has shown that the novel Optimal Driveline Control Strategy
(ODCS) for the rear-wheel-drive electric vehicles has some performance advantages
over the conventional drivelines which have been used successfully. For example, in
the project vehicle model, the new driveline with the ODCS has better driveability
performance than the model with the open or limited-slip differential. In addition,
research in this thesis has shown that the electric vehicle with the architecture of the
in-line motors can result in significant improvements in overall performance
compared with the electric vehicle with the in-wheel motors.
In Chapter 2, previous work on the improvement of vehicle dynamics and the direct
yaw control strategies are reviewed. With the rise of awareness of sustainable
development, interest in the electric vehicles has increased rapidly over recent years,
from the viewpoint of both industry and academic. Research and development efforts
have been focused on developing new concepts and low cost systems, but this has
proved difficult primarily because of high battery costs. Historically, the role of active
yaw control in the overall development of vehicle dynamics control technology has
commonly been focused. Active steering control and braking-based stability control
have actually played the main crucial roles in controlling the yaw moment of vehicles
and ensuring good safety, driveability and manoeuvrability. It is concluded that it has
been an area of rapidly changing technology over the past decade. But perhaps
more importantly, this is certain to continue over the next decade.
In Chapter 3, the project vehicle has been represented in detail, with the entire
architecture of the given electric vehicle with RWD being included, and the
components and the configuration of the new driveline also being introduced. The
capacity of the twin electric motors could fulfil the requirements of the racing vehicle.
The lightweight chassis and the wishbone suspension for the project vehicle were
149
described. Several sub-systems of the vehicle, for example the vehicle body, the
suspension system and the driveline, have been modelled in ADAMS, and the model
audit was carried out to ensure the reliability of the model. Also, the aerodynamic
effects were considered in this section. The driveline was able to produce the
variable driving torque at the wheels, so that it could represent the different driving
modes, such as the open differential and the torque vectoring control. The Magic
Formula tyre model was added to the vehicle model, and also used with the virtual
test rig model to examine characteristics of the tyre model in the intended behaviour
due to modifying the coefficients of the tyre model.
The analysis of the full vehicle model was presented in Chapter 4. A summary of the
simulation results was collected from these investigations to evaluate the overall
characteristics of the project model during the different analyses. First, the
characteristics of the suspension system have been determined by the kinematic
analysis in which the results can be compared with the laboratory data from
manufacturers to validate and refine the computer-based vehicle model.
Second, the dynamic analysis proved the model is a stable system with correct
degrees of freedom, the natural frequencies at each mode being calculated and the
animations clearly explaining the movement of the sprung and unsprung mass. Also
the influences of the natural frequencies at each mode were analysed. Finally, the
vehicle model ran for the characteristic driving manoeuvre - steady state cornering,
and implementing the control signal from the driver model with path following and
survey controller on the steering system was represented. Of most importance is
the fact that the characteristics of the project vehicle can be defined. Furthermore,
Figure 4-23 for the yaw rate gain against the vehicle speed indicated that the aim of
the further control strategy for active yaw control is to overcome all the variations,
and then to dominate the vehicle running at its particular characteristics.
In chapter 5, the vehicle architectures of in-wheel motors and in-line motors are
introduced, and the significant modifications of original suspension have been made
to integrate all components of in-wheel motors, such as the braking system, cooling
system and electric motors, into the constrained volume at the inside of the rim.
Furthermore, the un-sprung mass is obviously increased for implementing in-wheel
150
motors on a vehicle, and degrades ride and handling performance. Compared to the
architecture of the in-wheel motors vehicle, the in-line motors vehicle required the
minimum variations of the suspension system, thus it can retain the un-sprung
masses and reduce redesign costs.
Objective measurements were carried out by using several computing simulation

models in ADAMS, such as kinematic, dynamic and handling models. The aspects of
simulation have been examined in detail and it can be concluded that the overall
performance of ride comfort is degraded, and that the obvious impact of steering
behaviour emerges with testing the vehicle models in cornering and lane change
manoeuvres. The effects on safety and driveability require some variations in
suspension component detail and increases in damping levels to restore agility.
Overall, developing the electric vehicle in its next state the architecture of the in-line
motors vehicle can preserve the characteristics of a conventional vehicle with high
performance; also, the redesign costs and efforts can be economized by improving
the capacity of the electric motors and battery life. Moreover the benefits of individual
wheel motor control can be implemented by using in-line wheel motors for
substantial improvements in vehicle behaviour.
Based on the basic principles of torque vectoring control, an example was carried
out to explain how the mechanical torque vectoring control improves the vehicle
driveability. As stated, re-allocating the driving torque on the rear wheel results in
substantial influences on the control of the yaw moment. There are several
promising outcomes from this work listed below; these must be interpreted in the
context of the design approach used.
The Optimal Driveline Control Strategy (ODCS) for RWD EVs equipped with an in-
line motor at the rear wheels and the independent motor control has been designed
to enhance vehicle driveability and yaw stability. The ODCS involves three parts: the
desired dynamics behaviours, the secondary control, and the advanced torque
vectoring control (ATVC). The use of all levels has been introduced in detail, so that
the desired dynamics behaviours have been found to define the operating region and
to retain the vehicle linearity. In order to track the desired dynamics, the secondary
151
control has been designed so that it can modify the traction force to minimize the
yaw rate error and retain the vehicle speed. Since the inputs from the secondary
control should be applied to the electric motor through the ATVC, the ATVC has
been developed to map the secondary control inputs to the electric motors.
The simulation results with the proposed driving control have been compared to
those with the other two types of driving modes in order to verify the performance of
the proposed speed control and the proposed optimal torque vectoring algorithm. It
has been shown from simulation results that the vehicle driveability and yaw stability
can be significantly improved at the same time or individually compared to the other
driving modes. Driving control algorithms for alternative types of electric drive system,
such as a 4WD system with independently driven in-line motors, will be investigated
in the future.
A procedure for the optimal PID controller design and evaluation of torque-vectoring
controller for fully electric vehicles has been presented. The results, obtained for a
vehicle with in-line motors, demonstrate the effectiveness of torque-vectoring control
in tuning vehicle response. This is achieved through a set of reference dynamics
vehicle response in conditions of constant and variable vehicle velocities. For
implementation of this torque vectoring control system, the optimal PID controller
allows to evaluate the feed forward map of the control yaw moment, as a function of
measured and estimated quantities, for a given set of vehicle and tyre parameters.
The analysis of the optimal PID control criteria shows that energy-based cost
functions provide marginal benefit in the selection of the individual wheel torque
distribution. In contrast, performance index based on driving torque output allow a
smooth variation of the wheel torques for all achievable lateral accelerations and
yield.
Overall, the novel aspects of the work are: defining the characteristics and linearity of
the project vehicle using a novel consideration of yaw rate gain; the design and
development the Optimal Driveline Control Strategy (ODCS); the analysis and
modelling the ODCS in the vehicle and the comparison of the results with
conventional drivelines. The conclusions of the research are in-line with the original
152
aim and objectives. The overall performance benefits of equipping the project vehicle
with the new control strategy have been predicted to offer significant benefits in the
dynamic control over typical driving manoeuvres.
7.2. Future work
Future work could focus on other potential benefits of vehicle dynamic control, for
instance, the driveability of an electric vehicle with individual four-wheel control. Also,
the types of subjective assessments that are used to assess driveability, for example,
the stop-start condition, the feel of acceleration or deceleration, the control pedal
response and feel, and the body slip rate control could be investigated. It is clear
there exists a research opportunity to investigate how the objective numerical
measures are relative to the subjective assessments of expert assessors, and the
effects of different architectures on drivability.
Another potential benefit of the dynamic control for the electric vehicles which could
be investigated, is whether it is possible to downsize the motor and also reduce the
weight while maintaining acceleration ability in the limited times.
The simulation in this research suggests that the idea of using an optimal control
strategy in an electric vehicle will improve the driveability performance. In future work,
a four-wheel drive model could be taken into consideration and a more complex
control strategy for individual wheel control will be developed. The research could
also be focused on attempting to quantify both driving efficiency gains and
driveability improvements.
153
Bibliography
Ackermann J, J. G. (1995). Linear and Nonlinear Controller Design for Robust

Automotive Steering. IEEE Transactions on Control Sytems Technology,
Vol.3,No.1,1995, pp. 132-143.
al., J. C. (2005). Torque Vectoring Driveline: SUV-based Demonstrator and Practical

Actuation Technologies. SAE Internationa.
Andrs Eduardo Rojas Rojas, H. N. (2010). Comfort and Safety Enhancement of

Passenger Vehicles with In-Wheel Motors. SAE International.
Andrs Eduardo Rojas Rojas, H. N. (2011). Mechanical Design of In-Wheel Motor

Driven Vehicles with Torque-Vectoring. SAE International.

Driven Vehicles with Torque-Vectoring. SAE International.

Driven Vehicles with Torque-Vectoring,. SAE International.
Assadian, F. &. (2005). A comparison of yaw stability control strategies for the active
differential. In IEEE ISIE. Dubrovnik, Croatia.
Babik, B. (2006). Advanced Technology and Energy Strategies, General Motors.
Borner, M. &. (2006). Model-based detection of critical driving situations with fuzzy
logic decision making. Control Engineering Practice, 14(5),527.
C Ghike, T. S. (2009). Integrated control of wheel drivebrake torque for vehicle-

handling enhancement. Proceedings of the Institution of Mechanical
Engineers, Part D: Journal of Automobile Engineering.
Chan, C. C. (2001). Modern electric vehicle technology, Oxford University Press.
Cheli, F. M. (2005). A new control strategy for a semi-active differential (Part I),. 16th
IFAC World Congress.
Colombo, D. (2005). Active differential technology for yaw moment control. In sixth
all wheel drive congress. Austria: Gratz.
154
Cong Geng, L. M. (2009). Direct Yaw-Moment Control of an In-Wheel-Motored
Electric Vehicle Based on Body Slip Angle Fuzzy Observer. IEEE
TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO. 5,.
Edoardo Sabbioni, L. K. (2010). On the Impact of the Maximum Available Tire-Road

Friction Coefficient Awareness in a Brake-Based Torque Vectoring System.
SAE International.
Edoardo Sabbioni, L. K. (2010). On the Impact of the Maximum Available Tire-Road

Friction Coefficient Awareness in a Brake-Based Torque Vectoring System.
SAE International.
ExxonMobil. (2007). New Outlook for Energy: A View to 2030. Texas, ExxonMobil.
F. Assadian, M. H. (2008). Development of a Control Algorithm for an Active Limited

Slip Differential. Loughborough University / Society of Automotive Engineers
of Japan (JSAE).
F. Cheli, M. P. (2006). Development of a new control strategy for a semi-active

differential for a high-performance vehicle. Vehicle System Dynamics - VEH
SYST DYN 01/2006; 44:202-215. .
G. Kaiser, F. H. (2011). Torque Vectroing with a feedback and feed forward

controller - applied to a through the road vehicle. IEEE Intelligent Vehicle
Symposium.
Gaspar, P. S. (2005). Recongurable control structure to prevent the rollover of

heavy vehicles. Control Engineering Practice, 13(6), 699.
Gerd Kaiser, F. H. (2011). Torque Vectoring with a feedback and feed forward
controller - applied to a through the road hybrid electric vehicle. IEEE.
Gerhard, J. L.-C. (2005). Robust yaw control design with active differential and active
roll control systems. In 16th IFAC world congress. Prague, Czech Republic.
Gillespie, T. D. (1992). Fundamentals of Vehicle Dynamics. SAE International.
155
Gschel, B. (2008). Ausgewhlte Kapitel der Verbrennungskraftmaschinen:
Alektrifizierung Antrieb. Institute for Internal Combustion Engines and
Thermodynamics, Graz University of Technology.
Harty, M. B. (2004). Multibody Systems Approach to Vehicle Dynamics. Elseviers

Science and Technology Rights Department.
Hirsch, R. L. (2005). Peaking of World Oil Production: Impacts, Mitigation & Risk
Management.
Husain, I. (2003). Electric and hybrid vehicles; design fundamentals., CRC Press.
Ikushima, Y. a. (1995). "A Study on the Effects of the Active Yaw Moment Control,".
SAE Technical Paper 950303, 1995, doi:10.4271/950303.
IME. (2009). Brochure for Integrating Technologies for Low Carbon. Integrating
Technologies for Low Carbon, Norfolk, Institution of Mechanical Engineers.
Inoue, K. S. (2007). Effect of the Right-and-left Torque Vectoring System in Various

Types of Drivetrain. SAE International, 2007-01-3645.
IPIECA. (2004). Transportation and climate change: opportunities, challenges and

longterm strategies. AN IPECA workshop, Baltimore, USA.
J.C. Wheals, M. D. (2006). Design and Simulation of a Torque VectoringRear Axle.

SAE International.
Johann Willberger, M. A. (2010). Motor Selection Criteria and Potentials of Electrified

All Wheel Drive Concepts for Passenger Cars by Add-on Wheel Hub Motors
on the Rear Axle. SAE International.

on the Rear Axle. SAE International.

on the Rear Axle. SAE International .
156
John Feltman, J. G. (2006). Mechatronic Torque Vectoring System with Enhanced
Controllability for Augmenting the Vehicle Agility and Safety. SAE
International.
Jonathan C. Wheals, H. B. (2004). Torque Vectoring AWD Driveline: Design,

Simulation, Capabilities and Control. SAE International ISBN 0-7680-1319-4.
Jonathan C. Wheals, H. B. (2004). Torque Vectoring AWD Driveline:

Design,Simulation, Capabilities and Control. SAE International.
Juyong Kang, Y. k. (2012). Control Allocation based Optimal Torque Vectoring for
4WD Electric Vehicle. SAE International.
Kakalis L., C. F. (2010). n the impact of the maximum available tire-road friction
coefficient awareness in a brake-based torque vectoring system. . In SAE
International Congress and Exposition, SAE Paper 2010-01-0116.
Kang, J. k. (2012). Control Allocation based Optimal Torque Vectoring for 4WD
Electric Vehicle. SAE Technical Paper 2012-01-0246.
Kim, D.-H. J. (2007). "Optimal brake torque distribution for a four-wheel drive hybrid
electric vehicle stability enhancement." . Proc. IMechE, Part D: J Automobile
Engineering vol. 221: 1357-1366.
Kiumars Jalali, K. B. (2008). Design of an Advanced Traction Controller for an

Electric Vehicle Equipped with Four Direct Driven In-Wheel Motors. SAE
International.
Kiumars Jalali, T. U. (2012). Development of a Fuzzy Slip Control System for Electric
Vehicles with In-wheel Motors. SAE International.
Kohen, P. &. (2004). Active steeringthe BMW approach towards modern steering
technology. . In SAE Technical Paper No.2004-01-1105.
Kroppe, J. P. (2004). Dana Torque Vectoring Differential Dynamic Trak. SAE

International.
Larminie, J. a. (2003). Electric vehicle technology explained. West Sussex, England,

J. Wiley.
157
Lifu Li, Z. (2011). Study on Torque Vectoring Differential for Vehicle Stability Control
via Hardware Hardwarein-. IEEE.
M Milehins, C. C. (2010). Handling behaviour of a TTR hybrid electric vehicle with

independent rear wheel torque control. In: AVEC10.
Martyn Anderson, D. H. (2011). Unsprung Mass with In-Wheel Motors - Myths and
Realities. SAE International .
Milliken, W. a. (1995). Race car vehicle dynamics. Society of Automotive Engineers,

Inc .
Mokhiamar, O. &. (2002). Active wheel steering and yaw moment control
combination to maximize stability as well as responsiveness during quick lane
change for active vehicle handling safety. Proceedings of the Institution of
Mechanical Engineers Part D, Journal of Automobile Engineering, 216(1),
115124.
Motoki Shino, M. N. (2001). Yaw-moment control of electric vehicle for improving

handling and stability. Society of Automotive Engineers of Japan, Inc. and
Elsevier Science B.V.
Neudorfer, H., Binder, A., & Wicker, N. (2006). Analyse von Unterschiedlichen
Fahrzyklen fur den einsatz von Elektrofahrzeugen . E&I 123 no 7-8 p,352-360.
OHara, S. R. (2005). VEHICLE PATH OPTIMIZATION OF EMERGENCY LANE

CHANGE MANEUVERS FOR VEHICLE SIMULATION.
Pacejka, H. a. (1991). Shear force generation by pneumatic tyres in steady state

conditions: a review of modelling aspects. Vehicle System Dynamics, 20, pp.
121176, 1991.
PINTO, L. A. (2011). YAW MOTION CONTROL OF A VEHICLE. International Patent

Cooperation Treaty.
Qin Liu, G. K. (2011). Two-Degree-of-Freedom LPV Control for a through-the-Road

Hybrid Electric Vehicle via Torque Vectoring. IEEE.
158
Randall, B. P. (2009). Electric cars - are they really 'green'? Low-carbon vehicles
2009, London, Institution of Mechanical Engineers.
Resta, F. G. (2005). A new control strategy of a semi-active differential (Part II). the
16th IFAC WORLD CONGRESS.
Sawase, K. U. (2007). Effect of the Right-and-left Torque Vectoring System in

Various Types of Drivetrain. SAE Technical Paper 2007-01-3645, 2007,
doi:10.4271/2007-01-3645.
Sawase, T. K. (2012). Classification and analysis of electric Classification and

analysis of electricpowered. Proceedings of the Institution of Mechanical
Engineers, Part D: Journal of Automobile Engineering.
Sharma, S. K. (2006). Torque Vectoring Axle and Four Wheel Steering: A Simulation
Study of Two Yaw Moment Generation Mechanisms. SAE International.
Sienel, A. J. (1993). Robust Yaw and Damping of Cars with Front and Rear Wheel
Steering. IEEE Transactions on Control Systems Technology,
Vol.1,No.1,1993,pp.15-20.
Taehyun Shim, G. A. (n.d.). Autonomous vehicle collision avoidance system using

path planning and model-predictive-control-based active front steering and
wheel torque control. Proceedings of the Institution of Mechanical Engineers,
Part D: Journal of Automobile Engineering.
Taeyoung Lee, J. K. (2010). Integration of Longitudinal and Lateral Human Driver

Models for Evaluation of the Vehicle Active Safety Systems. SAE International.
van Zanten, A. T. (1995). VDC, the vehicle dynamics control system of bosch. In
SAE Technical Paper No. 95759.
159
van Zanten, A. T. (2000). Bosch ESP systems: 5 years of experience. In SAE
Technical Paper No. 2000-01-1633.
Vilaplana, M. A. (2005). Control of yaw rate and sideslip in 4-wheel steering cars with
actuator constraints. Lecture Notes in Computer Science, (vol. 3355, pp.201
222). Berlin: Springer.
Whitehead, A. (2012). IN-WHEEL ELECTRIC MOTORS. The Packaging and

Integration Challenges Senior Mechanical Systems Engineer Protean Electric
Ltd, UK, p. 15.
Wu, L. L. (2011). Study on Torque Vectoring Differential for Vehicle Stability Control
via Hardware-in-Loop Simulation. IEEE 978-1-61284-486-2/111.
YASA motor. (2014). Retrieved from http://www.yasamotors.com/technology
Zhang, Y. H. (2006). "Performance modelling and optimization of a novel multi mode

hybrid powertrain.". Trans. ASME, J. Mech. Design 128(1): 79-89.
160
Appendix
Appendix 1 Westfield vehicle model for kinematic analysis (part of the model)
!-------------------------- Default Units for Model ---------------------------!

defaults units & length = mm &angle = deg & force = newton mass = kg &time = sec
defaults units & coordinate_system_type = cartesian & orientation_type = body313
!------------------------ Default Attributes for Model ------------------------!
defaults attributes & inheritance = bottom_up &icon_visibility = on grid_visibility = off &
size_of_icons = 20.0 & spacing_for_grid = 1000.0
!------------------------------ Adams/View Model ------------------------------!
model create &model_name = Westfield_Front_Drive_Kin_rig
model attributes model_name = Westfield_Front_Drive_Kin_rig & size_of_icons = 20.0
view erase
-------------------------------- Materials ----------------------------------!
material create & material_name = .Westfield_Front_Drive_Kin_rig.steel & adams_id = 1
&youngs_modulus = 2.07E+005 &poissons_ratio = 0.29 & density = 7.801E-006
!-------------------------------- Rigid Parts ---------------------------------!
! Create parts and their dependent markers and graphics
!----------------------------------- ground -----------------------------------!
! ****** Ground Part ******
defaults model &part_name = grounddefaults coordinate_system &
default_coordinate_system = .Westfield_Front_Drive_Kin_rig.ground
! ****** Markers for current part ******
marker create &
marker_name = .Westfield_Front_Drive_Kin_rig.ground.m_wheel_centre_ground adams_id = 2 &
location = -1167.999, 663.971, 0.0 & orientation = 0.0d, 0.0d, 0.0d
marker attributes &marker_name = m_wheel_centre_ground &size_of_icons = 20.0 marker create
&marker_name
= .Westfield_Front_Drive_Kin_rig.ground.MARKER_57 & adams_id = 57 & location = -1269.8,
336.933, -96.75 &orientation = 186.7030013032d, 79.0487029613d, 0.0
marker create &marker_name
= .Westfield_Front_Drive_Kin_rig.ground.Body_FD_Lower_wish_rev_joint_maker & adams_id = 42
& location = -1026.439, 190.41, -135.4 & orientation = 269.7809583075d, 90.0d, 0.0d
marker create &marker_nam
= .Westfield_Front_Drive_Kin_rig.ground.MARKER_55 & adams_id = 55 &location = -1183.751,
279.455, 121.4 &orientation = 0.0d, 0.0d, 0.0d
marker create marker_name
= .Westfield_Front_Drive_Kin_rig.ground.MARKER_63 adams_id = 63 &
location = -1183.751, 279.455, 121.4 &orientation = 178.4831821016d, 136.0112734674d, 0.0d
= .Westfield_Front_Drive_Kin_rig.ground.MARKER_99 & adams_id = 99 &
location = -1277.559, 290.911, 74.56 &orientation = 270.1944385707d, 90.0d, 90.0d
= .Westfield_Front_Drive_Kin_rig.ground.MARKER_80 & adams_id = 80
location = -1167.999, 663.971, -50.0 & orientation = 0.0d, 0.0d, 0.0d
marker create & marker_name

= .Westfield_Front_Drive_Kin_rig.ground.ground_wheel_base &adams_id = 88 & location = -
1167.999, 663.971, -266.0 &orientation = 0.0d, 0.0d, 0.0d
= .Westfield_Front_Drive_Kin_rig.ground.MARKER_108 &adams_id = 108 &location = -1167.999,
663.971, -200.0 &orientation = 0.0d, 0.0d, 0.0d
part create rigid_body mass_properties part_name= .Westfield_Front_Drive_Kin_rig.ground
&material_type = .Westfield_Front_Drive_Kin_rig.steel
! ****** Points for current part ******
161
point create &point_name = .Westfield_Front_Drive_Kin_rig.ground.upper_wishbone_outer_ball_joint
& location = -1158.169, 551.78, 74.5point attributes &point_name = upper_wishbone_outer_ball_joint
& size_of_icons = 20.0
!
point create &point_name
= .Westfield_Front_Drive_Kin_rig.ground.upper_wishbone_inner_Rear_joint &location = -1085.433,
291.563, 74.5 point create &
point_name= .Westfield_Front_Drive_Kin_rig.ground.upper_wishbone_inner_Front_joint &location =
-1277.559, 290.911, 74.56
point attributes & point_name = upper_wishbone_inner_Front_joint &size_of_icons = 20.0 point
create &
point_name = .Westfield_Front_Drive_Kin_rig.ground.POINT_25 & location = -1050.0, 350.0, 0.0
point create & point_name = .Westfield_Front_Drive_Kin_rig.ground.lower_wishbone_outer_ball_joint
& location = -1177.704, 570.758, -135.4
point attributes point_name = lower_wishbone_outer_ball_joint &size_of_icons = 20.0
!----------------------------- FD_Upper_wishbone ------------------------------

defaults coordinate_system &default_coordinate_system = .Westfield_Front_Drive_Kin_rig.ground
part create rigid_body name_and_position &part_name
= .Westfield_Front_Drive_Kin_rig.FD_Upper_wishbone & adams_id = 8 &
location = 0.0, 0.0, 0.0 & orientation = 0.0d, 0.0d, 0.0d
part create rigid_body initial_velocity &
part_name = .Westfield_Front_Drive_Kin_rig.FD_Upper_wishbone &vx = -2.7778E+004
defaults coordinate_system & default_coordinate_system
= .Westfield_Front_Drive_Kin_rig.FD_Upper_wishbone

marker create &marker_name = .Westfield_Front_Drive_Kin_rig.FD_Upper_wishbone.MARKER_98
&adams_id = 98
location = -1277.559, 290.911, 74.56 &
orientation = 270.1944385707d, 90.0d, 90.0
marker create & marker_name = .Westfield_Front_Drive_Kin_rig.FD_Upper_wishbone.MARKER_35
adams_id = 35 &
location = -1085.433, 291.563, 74.56 &
orientation = 65.4081991848d, 0.0d, 0.0
!----------------------------- FD_Lower_wishbone ------------------------------!

defaults coordinate_system &default_coordinate_system = .Westfield_Front_Drive_Kin_rig.ground
part create rigid_body name_and_position & part_name
= .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone &
adams_id = 7 & location = 0.0, 0.0, 0.0 & orientation = 0.0d, 0.0d, 0.0d
part create rigid_body initial_velocity & part_name

= .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone &vx = -2.7778E+004
defaults coordinate_system &default_coordinate_system
= .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone
marker create & marker_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_30
adams_id = 30 &
location = -1177.704, 570.758, -135.4 &
orientation = 291.6877938567d, 0.0d, 0.0d
marker create &marker_name= .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_31
&adams_id = 31 &
location = -1026.439, 190.41, -135.4 &
orientation = 291.6877938567d, 0.0d, 0.0d
marker create &marker_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.cm &
adams_id = 43 &
location = -1179.1194902269, 380.878555804, -135.4 &
orientation = 179.2773770674d, 90.0000000007d, 89.9999999907d
162
marker create &marker_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_32
adams_id = 32 &
location = -1177.704, 570.758, -135.4 &
orientation = 291.6877938567d, 0.0d, 0.0d
marker create &
marker_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_33 &
adams_id = 33 &
location = -1334.311, 191.587, -135.4 orientation = 291.6877938567d, 0.0d, 0.0
marker create & marker_name

= .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.FD_Lower_wish_rev_joint_marker
&adams_id = 41 &
location = -1026.439, 190.41, -135.4 &
orientation = 269.7809583075d, 90.0d, 0.0d
!
marker create &

marker_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.lower_wishbone_ball_joint &
adams_id = 47 &
location = -1177.704, 570.758, -135.4 &
orientation = 0.0d, 0.0d, 0.0d
!
marker create &
marker_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_70 &
adams_id = 70 &
location = -1178.278, 486.142, -92.79 &
!
part create rigid_body mass_properties &
part_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone &
mass = 2.497681221 &
center_of_mass_marker = &
.Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.cm &
ixx = 5.130318723E+004 &
iyy = 3.158268078E+004 &
izz = 1.984539051E+004 &
ixy = 0.0 &
izx = 0.0 &
iyz = 0.0
! ****** Graphics for current part ******
geometry create shape link &

link_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.LINK_19 &
i_marker = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_30 &
j_marker = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_31 &
width = 40.0 &
depth = 20.0
!
geometry create shape link &

link_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.LINK_20 &
i_marker = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_32 &
j_marker = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone.MARKER_33 &
width = 40.0 &
depth = 20.0
part attributes &

part_name = .Westfield_Front_Drive_Kin_rig.FD_Lower_wishbone &
color = MAIZE &
name_visibility = off
163
!---------------------------------- upright -----------------------------------!
defaults coordinate_system &

default_coordinate_system = .Westfield_Front_Drive_Kin_rig.ground
!
part create rigid_body name_and_position &
part_name = .Westfield_Front_Drive_Kin_rig.upright &
adams_id = 4 &
location = 0.0, 0.0, 0.0 &
part_name = .Westfield_Front_Drive_Kin_rig.upright &
vx = -2.7778E+004
!
default_coordinate_system = .Westfield_Front_Drive_Kin_rig.upright
marker create &

marker_name = .Westfield_Front_Drive_Kin_rig.upright.MARKER_34 &
adams_id = 34 &
location = -1167.999, 663.971, 0.0 &
!
marker create &

marker_name = .Westfield_Front_Drive_Kin_rig.upright.m_upright_CG &
adams_id = 17 &
location = -1167.999, 561.269, 0.0 &
!
marker attributes &

marker_name = m_upright_CG &
size_of_icons = 20.0
!
marker create &
marker_name = .Westfield_Front_Drive_Kin_rig.upright.m_upright_graphics &
adams_id = 18 &
location = -1177.704, 570.758, -135.4 &
orientation = 45.8285909785d, 7.391038812d, 313.9325193345d
!
marker attributes &
marker_name = m_upright_graphics &
!
marker create &

marker_name = .Westfield_Front_Drive_Kin_rig.upright.m_upright_graphics2 &
adams_id = 19 &
location = -1298.093, 577.671, -49.847 &
orientation = 82.8141733619d, 69.1856027052d, 272.5651822746d
Appendix 2 Westfield model for Eigen solution ( part of model)
!-------------------------- Default Units for Model ---------------------------!

defaults units &
length = mm &
164
angle = deg &
force = newton &
mass = kg &
time = sec
!
defaults units &
coordinate_system_type = cartesian &
orientation_type = body313
!
!------------------------ Default Attributes for Model ------------------------!
defaults attributes &
inheritance = bottom_up &
icon_visibility = on &
grid_visibility = off &
size_of_icons = 20.0 &
spacing_for_grid = 1000.0
!
!------------------------------ Adams/View Model ------------------------------!
model create &
model_name = Westfield_full_vehicle_Eigen
!
model attributes &
model_name = .Westfield_full_vehicle_Eigen &
!
view erase
!-------------------------------- Data storage --------------------------------!
data_element create variable &
variable_name = .Westfield_full_vehicle_Eigen.YAW_RATE_VAR &
adams_id = 25136 &
initial_condition = 0.0 &
function = ""
!
data_element create spline &
spline_name = .Westfield_full_vehicle_Eigen.SPLINE_G686 &
adams_id = 1000 &
comments = "DAMPER SPLINES", "G686 WESTFIELD 3742078" &
x = -2000.0, -840.995, -800.995, -760.995, -720.995, -680.995, -660.995, &
-580.995, -520.995, -460.995, -420.995, -380.995, -340.995, -300.995, &
-270.995, -240.995, -228.995, -207.995, -192.887, -174.38, -161.916, &
-144.542, -128.301, -113.57, -98.84, -85.998, -71.268, -57.293, &
-43.696, -27.455, -15.746, 0.0, 15.746, 27.455, 38.408, 47.851, &
53.894, 64.092, 75.045, 83.354, 90.153, 100.728, 112.059, 122.257, &
130.944, 140.765, 152.096, 161.16, 167.959, 176.646, 184.2, 193.265, &
204.974, 214.974, 224.974, 234.974, 244.974, 254.974, 264.974, &
274.974, 284.974, 294.974, 304.974, 314.974, 380.0 &
y = 2000.0, 1186.0, 1156.0, 1126.0, 1096.0, 1066.0, 1036.0, 1006.0, 976.0, &
946.0, 916.0, 886.0, 856.0, 826.0, 796.0, 766.0, 736.0, 706.522, &
670.29, 634.058, 606.884, 570.652, 525.362, 498.188, 461.956, 434.782, &
389.493, 362.319, 317.029, 262.681, 226.449, 0.0, -280.797, -452.899, &
-625.0, -778.986, -878.623, -996.377, -1105.073, -1195.652, -1277.174, &
-1385.87, -1494.565, -1603.261, -1702.899, -1802.536, -1938.406, &
-2028.986, -2110.507, -2255.435, -2391.304, -2536.232, -2726.449, &
-2926.0, -3126.0, -3326.0, -3526.0, -3726.0, -3926.0, -4126.0, &
-4326.0, -4526.0, -4726.0, -4926.0, -8000.0 &
linear_extrapolate = no
!
data_element attributes &
data_element_name = .Westfield_full_vehicle_Eigen.SPLINE_G686 &
visibility = off
165
!
spline_name = .Westfield_full_vehicle_Eigen.SPLINE_West_1 &
adams_id = 2000 &
comments = "WESTFIELD 1" &
x = -1000.0, -273.24, -262.24, -251.24, -240.24, -229.24, -218.24, &
-207.24, -196.287, -182.689, -170.603, -157.006, -143.031, -130.189, &
-118.858, -104.128, -87.131, -78.067, -69.002, -58.048, -41.807, &
-29.343, -16.124, 0.0, 16.124, 35.764, 48.606, 60.315, 70.135, 80.333, &
90.153, 101.484, 110.549, 120.369, 131.7, 144.919, 157.383, 174.38, &
186.089, 197.797, 206.484, 216.0, 226.0, 236.0, 250.0, 267.0, 520.0 &
y = 2400.0, 750.0, 720.0, 690.0, 660.0, 630.0, 600.0, 588.768, 570.652, &
534.42, 507.246, 498.188, 461.956, 434.782, 416.667, 371.377, 335.145, &
317.029, 289.855, 262.681, 199.275, 153.985, 108.696, 0.0, -63.406, &
-217.391, -326.087, -434.783, -525.362, -625.0, -724.638, -824.276, &
-896.739, -969.203, -1068.841, -1186.594, -1277.174, -1422.102, &
-1530.797, -1630.435, -1702.899, -1772.0, -1870.0, -1970.0, -2100.0, &
-2200.0, -5000.0 &
!
data_element_name = .Westfield_full_vehicle_Eigen.SPLINE_West_1 &
visibility = off
!
spline_name = .Westfield_full_vehicle_Eigen.SPLINE_West_2 &
adams_id = 3000 &
comments = "WESTFIELD 2" &
x = -1000.0, -272.995, -251.995, -239.995, -228.995, -215.995, -201.952, &
-187.222, -170.603, -156.628, -142.653, -131.322, -118.103, -104.883, &
-92.041, -81.088, -71.268, -60.692, -50.117, -38.03, -25.944, -15.746, &
0.0, 15.746, 25.566, 36.52, 47.473, 53.894, 60.315, 68.624, 79.955, &
90.153, 99.973, 109.793, 119.613, 128.678, 137.743, 149.829, 159.65, &
170.603, 180.423, 192.51, 206.484, 212.484, 231.0, 251.0, 272.0, &
1000.0 &
y = 1448.5058131529, 601.0, 584.0, 551.0, 519.0, 498.188, 489.13, 461.956, &
434.782, 407.609, 380.435, 353.261, 326.087, 298.913, 271.739, &
235.507, 217.391, 190.217, 163.043, 135.869, 108.696, 90.58, 0.0, &
-36.232, -63.406, -90.58, -117.754, -135.87, -181.16, -217.391, &
-280.797, -344.203, -407.609, -471.015, -543.478, -606.884, -670.29, &
-751.812, -806.16, -878.623, -951.087, -1032.609, -1114.131, -1210.0, &
-1290.0, -1410.0, -1580.0, -6756.1418342369 &
!
data_element_name = .Westfield_full_vehicle_Eigen.SPLINE_West_2 &
visibility = off
!
spline_name = .Westfield_full_vehicle_Eigen.LEFT_TURN_SPLINE &
adams_id = 7000 &
x = 0.0, 8.333, 16.666, 21.5, 24.999, 33.332, 41.665, 49.998, 58.331, &
63.0, 66.664, 74.997, 83.33, 91.663, 99.996, 108.329, 116.662, &
124.995, 133.328, 141.661, 149.994, 158.327, 166.66, 174.993, 183.326, &
191.659, 199.992, 208.325, 216.658, 224.991, 233.324, 241.657, 249.99, &
258.323, 266.656, 274.989, 283.322, 291.655, 299.988, 308.321, &
316.654, 324.987, 333.32, 341.653, 349.986, 358.319, 366.652, 374.985, &
383.318, 391.651, 399.984, 408.317, 416.65, 424.983, 433.316, 441.649, &
449.982, 458.315, 466.648, 474.981, 483.314, 491.647 &
y = 0.0, 0.0, -1.0E-003, -2.7575744042E-003, -5.05E-003, -1.01E-002, &
166
-1.515E-002, -2.02E-002, -2.4997042955E-002, -2.7618766292E-002, &
-2.9065236397E-002, -2.9969280214E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, -3.03E-002, &
-3.03E-002, -3.03E-002, -3.03E-002 &
linear_extrapolate = yes
!
data_element_name = .Westfield_full_vehicle_Eigen.LEFT_TURN_SPLINE &
visibility = off
!--------------------------------- Materials ----------------------------------

material create &
material_name = .Westfield_full_vehicle_Eigen.steel &
adams_id = 1 &
youngs_modulus = 2.07E+005 &
poissons_ratio = 0.29 &
density = 7.801E-006
!-------------------------------- Rigid Parts ---------------------------------!
!
! Create parts and their dependent markers and graphics
!
!----------------------------------- ground -----------------------------------!
! ****** Ground Part ******
defaults model &
part_name = ground
default_coordinate_system = .Westfield_full_vehicle_Eigen.ground
marker create &
marker_name = .Westfield_full_vehicle_Eigen.ground.m_wheel_centre_ground &
adams_id = 2 &
location = -1167.999, 663.971, 0.0 &
!
marker attributes &
marker_name = .Westfield_full_vehicle_Eigen.ground.m_wheel_centre_ground &
!
marker create &
marker_name = .Westfield_full_vehicle_Eigen.ground.MARKER_150 &
adams_id = 150 &
location = 1163.586, 399.311, 419.149 &
!
marker create &
marker_name = .Westfield_full_vehicle_Eigen.ground.ground_wheel_base &
adams_id = 200 &
location = 1169.081, 663.971, -266.0 &
!
marker create &
167
adams_id = 292 &
location = 1163.586, -399.311, 419.149 &
!
marker create &
adams_id = 342 &
location = -1334.311, 191.587, -255.4 &
!
marker create &
adams_id = 369 &
location = 993.63, 134.002, -255.4 &
!
marker create &
adams_id = 404 &
location = -1269.8, 336.933, -33.12 &
!
marker create &
marker_name = .Westfield_full_vehicle_Eigen.ground.aero_drag_marker &
adams_id = 405 &
location = -1.00964, 0.0, -33.7444 &
!------------------------------- upright_drive --------------------------------!
default_coordinate_system = .Westfield_full_vehicle_Eigen.ground
!
part create rigid_body name_and_position &
part_name = .Westfield_full_vehicle_Eigen.upright_drive &
adams_id = 13 &
location = 0.0, 0.0, 0.0 &
!
vx = 0.0
!
default_coordinate_system = .Westfield_full_vehicle_Eigen.upright_drive
!
!
marker create &
marker_name = .Westfield_full_vehicle_Eigen.upright_drive.MARKER_86 &
adams_id = 86 &
location = -1158.169, 551.78, 74.56 &
orientation = 226.0112237762d, 169.6160562249d, 46.4838491772d
!
marker create &
marker_name = .Westfield_full_vehicle_Eigen.upright_drive.cm &
adams_id = 103 &
location = -1188.6920986753, 576.0213355289, -21.7272633248 &
orientation = 278.3869741337d, 153.2999544485d, 183.8203841601d
!
marker create &
168
adams_id = 87 &
location = -1167.999, 561.269, 0.0 &
orientation = 225.6447517964d, 174.2755531549d, 45.7879283968d
!
marker create &
adams_id = 88 &
location = -1298.093, 577.671, -49.847 &
orientation = 82.8141733619d, 69.1856027052d, 272.5651822746d
!
marker create &
adams_id = 89 &
location = -1167.999, 561.269, 0.0 &
!
marker create &
adams_id = 91 &
location = -1158.169, 551.78, 74.56 &
!
marker create &
adams_id = 93 &
location = -1177.704, 570.758, -135.4 &
!
marker create &
adams_id = 94 &
location = -1298.093, 577.671, -49.847 &
!
marker create &
adams_id = 504 &
location = -1167.999, 663.971, 0.0 &
!
part create rigid_body mass_properties &
mass = 7.027406137 &
center_of_mass_marker = .Westfield_full_vehicle_Eigen.upright_drive.cm &
ixx = 2.7196645587E+004 &
iyy = 2.4214601705E+004 &
izz = 1.1083130452E+004 &
ixy = 0.0 &
izx = 0.0 &
iyz = 0.0
!
! ****** Graphics for current part ******
!
geometry create shape cylinder &
cylinder_name = .Westfield_full_vehicle_Eigen.upright_drive.CYLINDER_41 &
adams_id = 41 &
center_marker = .Westfield_full_vehicle_Eigen.upright_drive.MARKER_86 &
angle_extent = 360.0 &
length = 75.8014750582 &
radius = 15.0 &
169
side_count_for_body = 20 &
segment_count_for_ends = 20
!
adams_id = 42 &
length = 136.0786101707 &
radius = 15.0 &
!
adams_id = 43 &
length = 140.2790000285 &
radius = 15.0 &
!
adams_id = 44 &
length = 102.702 &
radius = 15.0 &
!
part attributes &
color = YELLOW &
name_visibility = off
Appendix 3 Optimal driveline control strategy in ADAMS
!---------------------------- Function definitions ----------------------------!
constraint modify motion_generator &

motion_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.STEERING_MOTION &
function = "STEP(TIME,0,0,4,-80D)"
!
data_element modify variable &
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.VARIABLE_20000 &
function = "0.4"
!
function = "1.247E-9"
!
function = "573300"
!
170
function =
"VX(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.aero_drag_marker,.Westfield_PAT
H_FOLLOWING_CORNERING_TVD.vehicle_body.aero_drag_force,.Westfield_PATH_FOLLOWING
_CORNERING_TVD.vehicle_body.aero_drag_force)"
!
function =
"VY(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.aero_drag_marker,.Westfield_PAT
H_FOLLOWING_CORNERING_TVD.vehicle_body.aero_drag_force,.Westfield_PATH_FOLLOWING
_CORNERING_TVD.vehicle_body.aero_drag_force)"
!
function = "1000.0"
!
variable_name
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.GROUND_PLANE_VELOCITY &
function =
"((VX(.Westfield_PATH_FOLLOWING_CORNERING_TVD.vehicle_body.MARKER_9000,.Westfield_
PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100)**2+VY(.Westfield_PATH_FOLLOW
ING_CORNERING_TVD.vehicle_body.MARKER_9000,.Westfield_PATH_FOLLOWING_CORNERIN
G_TVD.ground.MARKER_100)**2)**0.5)/1000"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.PATH_LENGTH &
function = "DIF(.Westfield_PATH_FOLLOWING_CORNERING_TVD.DIFF_path_length)"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.DEMANDED_YAW_RATE
&
function = "-
VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.GROUND_PLANE_VELOCITY)*AKIS
PL(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.PATH_LENGTH),0,.Westfield_PAT
H_FOLLOWING_CORNERING_TVD.PATH_CURV_SPLINE)"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.BODY_SLIP_ANGLE &
function =
"ASIN((VY(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100,.Westfield_P
ATH_FOLLOWING_CORNERING_TVD.vehicle_body.MARKER_9000,.Westfield_PATH_FOLLOWIN
G_CORNERING_TVD.vehicle_body.MARKER_9000)/1000)/(VARVAL(.Westfield_PATH_FOLLOWIN
G_CORNERING_TVD.GROUND_PLANE_VELOCITY)+0.00001))"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.LAT_ACC &
function =
"ACCY(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100,.Westfield_PATH
_FOLLOWING_CORNERING_TVD.vehicle_body.MARKER_9000,.Westfield_PATH_FOLLOWING_C
ORNERING_TVD.vehicle_body.MARKER_9000)/1000"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.LONG_ACC &
function =
"ACCX(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100,.Westfield_PATH
ORNERING_TVD.vehicle_body.MARKER_9000)/1000"
!
171
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.CEN_ACC &
function = "-
(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.LAT_ACC)*COS(VARVAL(.Westfield
_PATH_FOLLOWING_CORNERING_TVD.BODY_SLIP_ANGLE)))+(VARVAL(.Westfield_PATH_FOL
LOWING_CORNERING_TVD.LONG_ACC)*SIN(VARVAL(.Westfield_PATH_FOLLOWING_CORNER
ING_TVD.BODY_SLIP_ANGLE)))"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.FRONT_AXLE_YAW_RATE
&
function = "-(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.CEN_ACC)-
WDTZ(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100,.Westfield_PATH
ORNERING_TVD.vehicle_body.MARKER_9000)*(DX(.Westfield_PATH_FOLLOWING_CORNERING
_TVD.vehicle_body.MARKER_9000,.Westfield_PATH_FOLLOWING_CORNERING_TVD.vehicle_bo
dy.MARKER_139,.Westfield_PATH_FOLLOWING_CORNERING_TVD.vehicle_body.MARKER_9000
)/1000))/(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.GROUND_PLANE_VELOCI
TY)+0.00001)"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.YAW_RATE_ERROR &
function =
"VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.DEMANDED_YAW_RATE)-
VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.FRONT_AXLE_YAW_RATE)"
!
variable_name
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.YAW_RATE_ERROR_INTEGARL &
function = "DIF(.Westfield_PATH_FOLLOWING_CORNERING_TVD.DIFF_YRE_integral)"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.YAW_RATE_GAIN &
function =
"VX(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100,.Westfield_PATH_F
OLLOWING_CORNERING_TVD.vehicle_body.MARKER_9000,.Westfield_PATH_FOLLOWING_CO
RNERING_TVD.vehicle_body.MARKER_9000)/2337.0"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.YAW_RATE_VAR &
function =
"WZ(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100,.Westfield_PATH_F
OLLOWING_CORNERING_TVD.vehicle_body.MARKER_9000)"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.PI_CONTROLLER &
function =
"VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.YAW_RATE_ERROR)*100000+VAR
VAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.YAW_RATE_ERROR_INTEGARL)*1000"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.DRIVE_TORQUE &
function =
"STEP(TIME,0.5,0,1,500*(abs(.Westfield_PATH_FOLLOWING_CORNERING_TVD.VEH_LON_VEL)-
VX(.Westfield_PATH_FOLLOWING_CORNERING_TVD.ground.MARKER_100,.Westfield_PATH_FO
LLOWING_CORNERING_TVD.vehicle_body.MARKER_9000,.Westfield_PATH_FOLLOWING_COR
NERING_TVD.vehicle_body.MARKER_9000)))"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.TVD_Pass &
172
function =
"IF(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.CEN_ACC):0,0,VARVAL(.Westfiel
d_PATH_FOLLOWING_CORNERING_TVD.DRIVE_TORQUE)-
VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.PI_CONTROLLER))"
!
variable_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.TVD_Drive &
function =
"IF(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.CEN_ACC):0,0,VARVAL(.Westfiel
d_PATH_FOLLOWING_CORNERING_TVD.DRIVE_TORQUE)+VARVAL(.Westfield_PATH_FOLLO
WING_CORNERING_TVD.PI_CONTROLLER))"
!
part modify equation differential_equation &
differential_equation_name
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.DIFF_path_length &
function =
"VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.GROUND_PLANE_VELOCITY)"
!
part modify equation differential_equation &
differential_equation_name
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.DIFF_YRE_integral &
function = "VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.YAW_RATE_ERROR)"
!
force modify direct single_component_force &
single_component_force_name
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.FD_damping_force &
function = "-
CUBSPL(VR(.Westfield_PATH_FOLLOWING_CORNERING_TVD.FD_Lower_damper_drive.MARKE
R_56,.Westfield_PATH_FOLLOWING_CORNERING_TVD.FD_Upper_damper.MARKER_55),0,.West
field_PATH_FOLLOWING_CORNERING_TVD.SPLINE_G686)"
!
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.FP_damping_force &
function = "-
CUBSPL(VR(.Westfield_PATH_FOLLOWING_CORNERING_TVD.FP_upper_damper_pass.MARKE
R_124,.Westfield_PATH_FOLLOWING_CORNERING_TVD.FP_Lower_damper_pass.MARKER_125
),0,.Westfield_PATH_FOLLOWING_CORNERING_TVD.SPLINE_G686)"
!
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.RP_damping_force &
function = "-
CUBSPL(VR(.Westfield_PATH_FOLLOWING_CORNERING_TVD.RP_lower_damper.MARKER_366,
.Westfield_PATH_FOLLOWING_CORNERING_TVD.RP_lower_damper.MARKER_296),0,.Westfield_
PATH_FOLLOWING_CORNERING_TVD.SPLINE_G686)"
!
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.RD_damping_force &
function = "-
CUBSPL(VR(.Westfield_PATH_FOLLOWING_CORNERING_TVD.RD_lower_damper.MARKER_154,
.Westfield_PATH_FOLLOWING_CORNERING_TVD.RD_Upper_damper.MARKER_153),0,.Westfield
_PATH_FOLLOWING_CORNERING_TVD.SPLINE_G686)"
!
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.SFORCE_10 &
function =
"SIGN(1,VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.VARIABLE_23000))*((VARV
173
AL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.VARIABLE_20000)*VARVAL(.Westfield_PA
TH_FOLLOWING_CORNERING_TVD.VARIABLE_21000)*VARVAL(.Westfield_PATH_FOLLOWING
_CORNERING_TVD.VARIABLE_22000)*(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_T
VD.VARIABLE_23000)**2))/(2*VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.VARIA
BLE_25000)+0.1E-20))"
!
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.RD_DRIVE_TORQUE &
function = "VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.DRIVE_TORQUE)"
!
single_component_force_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.RD_TVD &
function = "IF((VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.CEN_ACC)-
2.943):0,0,STEP(TIME,0,0,4.5,1)*", &
"(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.DRIVE_TORQUE)", &
"-VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.PI_CONTROLLER)))"
!
single_component_force_name = .Westfield_PATH_FOLLOWING_CORNERING_TVD.RP_TVD &
function = "IF((VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.CEN_ACC)-
2.943):0,0,STEP(TIME,0,0,4.5,1)*", &
"(VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.DRIVE_TORQUE)", &
"+VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.PI_CONTROLLER)))"
!
= .Westfield_PATH_FOLLOWING_CORNERING_TVD.RP_DRIVE_TORQUE &
function = "VARVAL(.Westfield_PATH_FOLLOWING_CORNERING_TVD.DRIVE_TORQUE)"
Appendix 4 Optimised Cosine path curvature for double lane change
Optimised Cosine
Max 0.403298928
0.012378754
Instant Centre vehicle
frame
Xcv curvature Ay(g)
0 0 0
0.5 -0.000417955 -0.013620573
1 -0.001615372 -0.052642757
1.5 -0.003430534 -0.11179639
2 -0.005618293 -0.183092446
2.5 -0.00788318 -0.256901999
3 -0.00991931 -0.323256658
3.5 -0.011451691 -0.373194859
4 -0.012273368 -0.399972172
4.5 -0.012375451 -0.403298928
5 -0.012349041 -0.402438268
5.5 -0.012296278 -0.400718784
6 -0.012217274 -0.398144146
6.5 -0.012112198 -0.394719849
7 -0.011981273 -0.390453199
174
7.5 -0.01182478 -0.385353303
8 -0.011643052 -0.379431043
8.5 -0.011436477 -0.372699058
9 -0.011205497 -0.365171715
9.5 -0.010950603 -0.356865077
10 -0.01067234 -0.34779687
10.5 -0.010371302 -0.337986448
11 -0.010048131 -0.327454746
11.5 -0.009703516 -0.316224239
12 -0.009338194 -0.304318893
12.5 -0.008952944 -0.291764115
13 -0.008548587 -0.278586698
13.5 -0.008125988 -0.264814763
14 -0.007686047 -0.2504777
14.5 -0.007229704 -0.235606105
15 -0.006757932 -0.220231715
15.5 -0.006271739 -0.204387339
16 -0.005772161 -0.18810679
16.5 -0.005260265 -0.171424812
17 -0.004737144 -0.154377005
17.5 -0.004203913 -0.136999749
18 -0.003661711 -0.119330129
18.5 -0.003111695 -0.101405852
19 -0.002555038 -0.08326517
19.5 -0.001992928 -0.064946797
20 -0.001426566 -0.046489823
20.5 -0.000857159 -0.027933638
21 -0.000285923 -0.009317841
21.5 0.000285923 0.009317841
22 0.000857159 0.027933638
22.5 0.001426566 0.046489823
23 0.001992928 0.064946797
23.5 0.002555038 0.08326517
24 0.003111695 0.101405852
24.5 0.003661711 0.119330129
25 0.004203913 0.136999749
25.5 0.004737144 0.154377005
26 0.005260265 0.171424812
26.5 0.005772161 0.18810679
27 0.006271739 0.204387339
27.5 0.006757932 0.220231715
28 0.007229704 0.235606105
28.5 0.007686047 0.2504777
29 0.008125988 0.264814763
29.5 0.008548587 0.278586698
30 0.008952944 0.291764115
30.5 0.009338194 0.304318893
31 0.009703516 0.316224239
31.5 0.010048131 0.327454746
32 0.010371302 0.337986448
32.5 0.01067234 0.34779687
33 0.010950603 0.356865077
33.5 0.011205497 0.365171715
34 0.011436477 0.372699058
34.5 0.011643052 0.379431043
35 0.01182478 0.385353303
35.5 0.011981273 0.390453199
36 0.012112198 0.394719849
175
36.5 0.012217274 0.398144146
37 0.012296278 0.400718784
37.5 0.012349041 0.402438268
38 0.012375451 0.403298928
38.5 0.012273368 0.399972172
39 0.011451691 0.373194859
39.5 0.00991931 0.323256658
40 0.00788318 0.256901999
40.5 0.005618293 0.183092446
41 0.003430534 0.11179639
41.5 0.001615372 0.052642757
42 0.000417955 0.013620573
42.5 0 0
43 0 0
43.5 0 0
44 0 0
44.5 0 0
45 0 0
45.5 0 0
46 0 0
46.5 0 0
47 0.000417955 0.013620573
47.5 0.001615372 0.052642757
48 0.003430534 0.11179639
48.5 0.005618293 0.183092446
49 0.00788318 0.256901999
49.5 0.00991931 0.323256658
50 0.011451691 0.373194859
50.5 0.012273368 0.399972172
51 0.012375451 0.403298928
51.5 0.012349041 0.402438268
52 0.012296278 0.400718784
52.5 0.012217274 0.398144146
53 0.012112198 0.394719849
53.5 0.011981273 0.390453199
54 0.01182478 0.385353303
54.5 0.011643052 0.379431043
55 0.011436477 0.372699058
55.5 0.011205497 0.365171715
56 0.010950603 0.356865077
56.5 0.01067234 0.34779687
57 0.010371302 0.337986448
57.5 0.010048131 0.327454746
58 0.009703516 0.316224239
58.5 0.009338194 0.304318893
59 0.008952944 0.291764115
59.5 0.008548587 0.278586698
60 0.008125988 0.264814763
60.5 0.007686047 0.2504777
61 0.007229704 0.235606105
61.5 0.006757932 0.220231715
62 0.006271739 0.204387339
62.5 0.005772161 0.18810679
63 0.005260265 0.171424812
63.5 0.004737144 0.154377005
64 0.004203913 0.136999749
64.5 0.003661711 0.119330129
65 0.003111695 0.101405852
176
65.5 0.002555038 0.08326517
66 0.001992928 0.064946797
66.5 0.001426566 0.046489823
67 0.000857159 0.027933638
67.5 0.000285923 0.009317841
68 -0.000285923 -0.009317841
68.5 -0.000857159 -0.027933638
69 -0.001426566 -0.046489823
69.5 -0.001992928 -0.064946797
70 -0.002555038 -0.08326517
70.5 -0.003111695 -0.101405852
71 -0.003661711 -0.119330129
71.5 -0.004203913 -0.136999749
72 -0.004737144 -0.154377005
72.5 -0.005260265 -0.171424812
73 -0.005772161 -0.18810679
73.5 -0.006271739 -0.204387339
74 -0.006757932 -0.220231715
74.5 -0.007229704 -0.235606105
75 -0.007686047 -0.2504777
75.5 -0.008125988 -0.264814763
76 -0.008548587 -0.278586698
76.5 -0.008952944 -0.291764115
77 -0.009338194 -0.304318893
77.5 -0.009703516 -0.316224239
78 -0.010048131 -0.327454746
78.5 -0.010371302 -0.337986448
79 -0.01067234 -0.34779687
79.5 -0.010950603 -0.356865077
80 -0.011205497 -0.365171715
80.5 -0.011436477 -0.372699058
81 -0.011643052 -0.379431043
81.5 -0.01182478 -0.385353303
82 -0.011981273 -0.390453199
82.5 -0.012112198 -0.394719849
83 -0.012217274 -0.398144146
83.5 -0.012296278 -0.400718784
84 -0.012349041 -0.402438268
84.5 -0.012375451 -0.403298928
85 -0.012273368 -0.399972172
85.5 -0.011451691 -0.373194859
86 -0.00991931 -0.323256658
86.5 -0.00788318 -0.256901999
87 -0.005618293 -0.183092446
87.5 -0.003430534 -0.11179639
88 -0.001615372 -0.052642757
88.5 -0.000417955 -0.013620573
89 0 0
89.5 0 0
90 0 0
90.5 0 0
91 0 0
91.5 0 0
92 0 0
92.5 0 0
93 0 0
93.5 0 0
94 0 0
177
94.5 0 0
95 0 0
95.5 0 0
96 0 0
96.5 0 0
97 0 0
97.5 0 0
98 0 0
98.5 0 0
99 0 0
99.5 0 0
100 0 0
Appendix 5 Tables of the front and rear suspension hard points

Table 0-1 Front suspension hard points
Table 0-2 Rear suspension hard points
178
Appendix 5 Measurement outputs from the front and rear suspension system
All Figures show the outputs for the front suspension characteristics based on the
quarter car model that has been built in the previous section.
Figure 1 Half Track Change Vs Bump Movement (Front)
179
Figure 2 Wheel Recession Vs Bump Movement (Front)
Figure 3 Steering Axis Inclination Vs Bump Movement (Front)
Figure 4 Ground Level Offset Vs Bump Movement (Front)
180
Figure 5 Castor Angle Vs Bump Movement (Front)
Figure 6 Suspension Trail Vs Bump Movement (Front)
Figure 7 Steer Angle Vs Bump Movement (Front)
181
Figures show the outputs for the rear suspension characteristics based on the
quarter car model that has been built in the previous section.
Figure 9 Half Track Change Vs Bump Movement (Rear)
Figure 10 Wheel Recession Vs Bump Movement (Rear)
182
Figure 11 Steer Axis Inclination Vs Bump Movement (Rear)
Figure 12 Ground Level Offset Vs Bump Movement (Rear)
Figure 13 Castor Angle Vs Bump Movement (Rear)
183
Figure 14 Suspension Trail Vs Bump Movement (Rear)
Figure 15 Camber Angle Vs Bump Movement (Rear)
Figure 16 Steer Angle Vs Bump Movement (Rear)
184

Guo 2015

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Guo 2015

Caricato da

Copyright:

Formati disponibili

Design and modelling of optimal

driveline control strategy for an electric

CURVE is the Institutional Repository for Coventry University

A thesis submitted in partial fulfilment of the

requirements of the Coventry University

for the degree of Doctor of Philosophy

Index for Figures ....................................................................................................... IV

Index for tables ........................................................................................................ VIII

1.1 Pollution and fuel consumption ..................................................................... 1

1.2 Developing electric vehicles .......................................................................... 1

1.3 Aim and objectives ........................................................................................ 3

2 Literature Review ................................................................................................ 6

2.1 Introduction ................................................................................................... 6

2.2 Electric Vehicle.............................................................................................. 6

2.3 Drivetrain Designs ......................................................................................... 7

2.3.1 Conventional mechanical drivetrains ...................................................... 7

2.4.1 Direct Yaw Control ................................................................................ 11

2.5.1 Driver Model ......................................................................................... 22

3. Modelling Electric Vehicle in ADAMS ................................................................ 29

3.1 Introduction ................................................................................................. 29

3.2 Project Vehicle ............................................................................................ 29

3.3.1 Vehicle Body......................................................................................... 34

3.4.1 Modelling virtual tyre rig ........................................................................ 47

4 Measurement and analysis of the virtual model ................................................ 53

4.1 Introduction ................................................................................................. 53

4.2 Kinematic analysis ...................................................................................... 53

4.2.1 Quarter vehicle modelling ..................................................................... 54

4.3.1 Total degrees of freedom...................................................................... 61

4.4.1 Driver behaviour modelling a path following controller ..................... 68

4.5.1 Steady state cornering behaviour ......................................................... 77

5. Comparison of architecture for electric vehicle ................................................. 83

5.1 Introduction ................................................................................................. 83

5.2 Architecture of in-wheel motors vehicle ....................................................... 83

5.3 Modelling and validating of in-wheel motors vehicle ................................... 86

5.5 Simulation Results ...................................................................................... 91

5.3.1 Analysis of ride comfort ........................................................................ 91

6 Torque vectoring system with in-line motors ................................................... 109

6.1 Introduction ............................................................................................... 109

6.2 Basic principles ......................................................................................... 110

6.3 Optimal Driveline Control Strategy (ODCS) .............................................. 114

6.3.1 Definitions of Desired Dynamics Behaviour ........................................ 116

6.4.1 Steady state cornering ........................................................................ 136

7. Conclusions and future work ........................................................................... 149

7.1. Summary and conclusions ........................................................................ 149

7.2. Future work ............................................................................................... 153

Bibliography ........................................................................................................... 154

Appendix ................................................................................................................ 161

Figure 2-1 Acceleration cornering performance ...................................................... 13

Table 3-1 Front suspension hard points ................................................................. 178

ALSD Active Limited-Slip Differential

I am deeply indebted to my director of study Professor Mike Blundell, who gave me

1.1 Pollution and fuel consumption

1.2 Developing electric vehicles

1.3 Aim and objectives

2.2 Electric Vehicle

Based on the global requirement to reduce emissions, both political and

2.3 Drivetrain Designs

2.3.1 Conventional mechanical drivetrains

2.3.2 Drivetrains for electric vehicle

Current electric vehicle research is investigating different powertrain configurations,

Figure 2-1 Possible EV configurations (Ehsani, Gao et al. 2004)

Modification of the vehicle dynamics can also be achieved by controlling the

2.4.1 Direct Yaw Control

Due to unusual external conditions there can be unexpected dangerous behaviours